up wiki land

This commit is contained in:
Tykayn 2025-08-22 18:19:20 +02:00 committed by tykayn
parent 391a212034
commit e533c273b2
10 changed files with 1116 additions and 182 deletions

View file

@ -0,0 +1,4 @@
{
"last_updated": "2025-08-22T18:13:20.641943",
"recent_changes": []
}

View file

@ -8,6 +8,112 @@ use Symfony\Component\Routing\Annotation\Route;
class WikiController extends AbstractController
{
/**
* Detects incorrect heading hierarchies in a list of sections
* For example, h4 directly under h2 without h3 in between
*
* @param array $sections List of sections with 'level' and 'title' keys
* @return array List of section indices with hierarchy errors
*/
private function detectHeadingHierarchyErrors(array $sections): array
{
$errors = [];
$lastLevel = 0;
foreach ($sections as $index => $section) {
$currentLevel = isset($section['level']) ? (int)$section['level'] : 0;
// Skip if level is not set or is 0
if ($currentLevel === 0) {
continue;
}
// If this is the first section, just record its level
if ($lastLevel === 0) {
$lastLevel = $currentLevel;
continue;
}
// Check if the level jump is more than 1
// For example, h2 -> h4 (skipping h3)
if ($currentLevel > $lastLevel + 1) {
$errors[] = $index;
}
$lastLevel = $currentLevel;
}
return $errors;
}
#[Route('/wiki/recent-changes', name: 'app_admin_wiki_recent_changes')]
public function recentChanges(): Response
{
$recentChangesFile = $this->getParameter('kernel.project_dir') . '/wiki_compare/recent_changes.json';
// Initialize arrays
$recentChanges = [];
$lastUpdated = null;
// Check if the recent changes file exists and load it
if (file_exists($recentChangesFile)) {
$recentChangesData = json_decode(file_get_contents($recentChangesFile), true);
if (isset($recentChangesData['recent_changes']) && is_array($recentChangesData['recent_changes'])) {
$recentChanges = $recentChangesData['recent_changes'];
$lastUpdated = isset($recentChangesData['last_updated']) ? $recentChangesData['last_updated'] : null;
}
// Check if the data is older than 1 hour
if ($lastUpdated) {
$lastUpdatedTime = new \DateTime($lastUpdated);
$now = new \DateTime();
$diff = $now->diff($lastUpdatedTime);
// If older than 1 hour, refresh the data
if ($diff->h >= 1 || $diff->days > 0) {
$this->refreshRecentChangesData();
return $this->redirectToRoute('app_admin_wiki_recent_changes');
}
}
} else {
// If the file doesn't exist, try to create it by running the script
$this->refreshRecentChangesData();
// Check if the file was created
if (file_exists($recentChangesFile)) {
return $this->redirectToRoute('app_admin_wiki_recent_changes');
} else {
$this->addFlash('error', 'Impossible de générer le fichier des changements récents.');
}
}
return $this->render('admin/wiki_recent_changes.html.twig', [
'recent_changes' => $recentChanges,
'last_updated' => $lastUpdated
]);
}
/**
* Refresh the recent changes data by running the fetch_recent_changes.py script
*/
private function refreshRecentChangesData(): void
{
try {
$scriptPath = $this->getParameter('kernel.project_dir') . '/wiki_compare/fetch_recent_changes.py';
if (file_exists($scriptPath)) {
exec('python3 ' . $scriptPath . ' --force 2>&1', $output, $returnCode);
if ($returnCode !== 0) {
$this->addFlash('warning', 'Impossible de mettre à jour les changements récents. Erreur: ' . implode("\n", $output));
}
} else {
$this->addFlash('error', 'Le script fetch_recent_changes.py n\'existe pas.');
}
} catch (\Exception $e) {
$this->addFlash('error', 'Erreur lors de l\'exécution du script: ' . $e->getMessage());
}
}
#[Route('/wiki/missing-translations', name: 'app_admin_wiki_missing_translations')]
public function missingTranslations(): Response
{
@ -664,11 +770,149 @@ class WikiController extends AbstractController
$mediaComparison['fr_only_count'] = count($frOnlyImages);
}
// Get link comparison data
$linkComparison = $page['link_comparison'] ?? null;
// Sort links alphabetically by URL if link comparison exists
if ($linkComparison) {
// Sort English-only links
if (isset($linkComparison['en_only']) && is_array($linkComparison['en_only'])) {
usort($linkComparison['en_only'], function($a, $b) {
return strcmp($a['href'], $b['href']);
});
}
// Sort French-only links
if (isset($linkComparison['fr_only']) && is_array($linkComparison['fr_only'])) {
usort($linkComparison['fr_only'], function($a, $b) {
return strcmp($a['href'], $b['href']);
});
}
// Sort common links
if (isset($linkComparison['common']) && is_array($linkComparison['common'])) {
usort($linkComparison['common'], function($a, $b) {
return strcmp($a['en']['href'], $b['en']['href']);
});
}
}
// Get section comparison data and filter out "Contents" sections
$sectionComparison = $page['section_comparison'] ?? null;
// Filter out "Contents" sections if section comparison exists
if ($sectionComparison) {
// Filter common sections
if (isset($sectionComparison['common']) && is_array($sectionComparison['common'])) {
$sectionComparison['common'] = array_filter($sectionComparison['common'], function($section) {
// Skip if either English or French title is "Contents"
return !($section['en']['title'] === 'Contents' || $section['fr']['title'] === 'Sommaire');
});
// Re-index array
$sectionComparison['common'] = array_values($sectionComparison['common']);
}
// Filter English-only sections
if (isset($sectionComparison['en_only']) && is_array($sectionComparison['en_only'])) {
$sectionComparison['en_only'] = array_filter($sectionComparison['en_only'], function($section) {
return $section['title'] !== 'Contents';
});
// Re-index array
$sectionComparison['en_only'] = array_values($sectionComparison['en_only']);
}
// Filter French-only sections
if (isset($sectionComparison['fr_only']) && is_array($sectionComparison['fr_only'])) {
$sectionComparison['fr_only'] = array_filter($sectionComparison['fr_only'], function($section) {
return $section['title'] !== 'Sommaire';
});
// Re-index array
$sectionComparison['fr_only'] = array_values($sectionComparison['fr_only']);
}
}
// Calculate adjusted section counts (excluding "Contents" sections)
$enSectionCount = $enPage['sections'];
$frSectionCount = $frPage['sections'];
// Adjust section counts if we have section comparison data
if ($sectionComparison) {
// Count how many "Contents" sections were filtered out
$contentsFilteredCount = 0;
// Check common sections that were filtered
if (isset($page['section_comparison']['common']) && is_array($page['section_comparison']['common'])) {
foreach ($page['section_comparison']['common'] as $section) {
if ($section['en']['title'] === 'Contents' || $section['fr']['title'] === 'Sommaire') {
$contentsFilteredCount++;
}
}
}
// Check English-only sections that were filtered
if (isset($page['section_comparison']['en_only']) && is_array($page['section_comparison']['en_only'])) {
foreach ($page['section_comparison']['en_only'] as $section) {
if ($section['title'] === 'Contents') {
$contentsFilteredCount++;
}
}
}
// Check French-only sections that were filtered
if (isset($page['section_comparison']['fr_only']) && is_array($page['section_comparison']['fr_only'])) {
foreach ($page['section_comparison']['fr_only'] as $section) {
if ($section['title'] === 'Sommaire') {
$contentsFilteredCount++;
}
}
}
// Adjust section counts
$enSectionCount -= $contentsFilteredCount;
$frSectionCount -= $contentsFilteredCount;
}
// Check for incorrect heading hierarchies
$enHierarchyErrors = [];
$frHierarchyErrors = [];
// Check English sections
if (isset($sectionComparison['en_only']) && is_array($sectionComparison['en_only'])) {
$enHierarchyErrors = $this->detectHeadingHierarchyErrors($sectionComparison['en_only']);
}
// Also check common sections (English side)
if (isset($sectionComparison['common']) && is_array($sectionComparison['common'])) {
$commonEnSections = array_map(function($section) {
return $section['en'];
}, $sectionComparison['common']);
$enHierarchyErrors = array_merge($enHierarchyErrors, $this->detectHeadingHierarchyErrors($commonEnSections));
}
// Check French sections
if (isset($sectionComparison['fr_only']) && is_array($sectionComparison['fr_only'])) {
$frHierarchyErrors = $this->detectHeadingHierarchyErrors($sectionComparison['fr_only']);
}
// Also check common sections (French side)
if (isset($sectionComparison['common']) && is_array($sectionComparison['common'])) {
$commonFrSections = array_map(function($section) {
return $section['fr'];
}, $sectionComparison['common']);
$frHierarchyErrors = array_merge($frHierarchyErrors, $this->detectHeadingHierarchyErrors($commonFrSections));
}
$detailedComparison = [
'section_comparison' => $page['section_comparison'] ?? null,
'link_comparison' => $page['link_comparison'] ?? null,
'section_comparison' => $sectionComparison,
'link_comparison' => $linkComparison,
'media_comparison' => $mediaComparison,
'category_comparison' => $page['category_comparison'] ?? null
'category_comparison' => $page['category_comparison'] ?? null,
'adjusted_en_section_count' => $enSectionCount,
'adjusted_fr_section_count' => $frSectionCount,
'en_hierarchy_errors' => $enHierarchyErrors,
'fr_hierarchy_errors' => $frHierarchyErrors
];
$mediaDiff = $page['media_diff'] ?? 0;
@ -841,6 +1085,15 @@ class WikiController extends AbstractController
}
}
// Ensure page URLs are strings to prevent array to string conversion errors
if ($frPage && isset($frPage['url']) && is_array($frPage['url'])) {
$frPage['url'] = json_encode($frPage['url']);
}
if ($enPage && isset($enPage['url']) && is_array($enPage['url'])) {
$enPage['url'] = json_encode($enPage['url']);
}
return $this->render('admin/wiki_compare.html.twig', [
'key' => $key,
'en_page' => $enPage,

View file

@ -41,6 +41,11 @@
<i class="bi bi-people"></i> Groupes OSM-FR
</a>
</li>
<li class="nav-item">
<a class="nav-link {% if app.request.get('_route') == 'app_admin_wiki_recent_changes' %}active{% endif %}" href="{{ path('app_admin_wiki_recent_changes') }}">
<i class="bi bi-clock-history"></i> Changements récents
</a>
</li>
</ul>
</div>
</div>

View file

@ -58,7 +58,13 @@
<div class="card h-100">
<div class="card-header bg-primary text-white">
<h3>Sections en anglais</h3>
<span class="badge bg-light text-dark">{{ en_page.sections }} sections</span>
<span class="badge bg-light text-dark">
{% if detailed_comparison.adjusted_en_section_count is defined %}
{{ detailed_comparison.adjusted_en_section_count }} sections ({{ en_page.sections }} avec "Contents")
{% else %}
{{ en_page.sections }} sections
{% endif %}
</span>
</div>
<div class="card-body">
<h4>Sections alignées par hiérarchie</h4>
@ -67,6 +73,9 @@
<li class="list-group-item title-level-{{ section.en.level }}">
{# <span class="badge bg-secondary">h{{ section.en.level }}</span> #}
{{ section.en.title }}
{% if detailed_comparison.en_hierarchy_errors is defined and loop.index0 in detailed_comparison.en_hierarchy_errors %}
<span class="badge bg-danger ms-2" title="Hiérarchie incorrecte">!</span>
{% endif %}
</li>
{% endfor %}
@ -74,6 +83,9 @@
<li class="list-group-item list-group-item-warning title-level-{{ section.level }}">
{# <span class="badge bg-secondary">h{{ section.level }}</span> #}
{{ section.title }}
{% if detailed_comparison.en_hierarchy_errors is defined and (detailed_comparison.section_comparison.common|length + loop.index0) in detailed_comparison.en_hierarchy_errors %}
<span class="badge bg-danger ms-2" title="Hiérarchie incorrecte">!</span>
{% endif %}
</li>
{% endfor %}
</ul>
@ -84,7 +96,13 @@
<div class="card h-100">
<div class="card-header bg-info text-white">
<h3>Sections en français</h3>
<span class="badge bg-light text-dark">{{ fr_page.sections }} sections</span>
<span class="badge bg-light text-dark">
{% if detailed_comparison.adjusted_fr_section_count is defined %}
{{ detailed_comparison.adjusted_fr_section_count }} sections ({{ fr_page.sections }} avec "Sommaire")
{% else %}
{{ fr_page.sections }} sections
{% endif %}
</span>
</div>
<div class="card-body">
<h4>Sections alignées par hiérarchie</h4>
@ -93,6 +111,9 @@
<li class="list-group-item title-level-{{ section.fr.level }}">
{# <span class="badge bg-secondary">h{{ section.fr.level }}</span> #}
{{ section.fr.title }}
{% if detailed_comparison.fr_hierarchy_errors is defined and loop.index0 in detailed_comparison.fr_hierarchy_errors %}
<span class="badge bg-danger ms-2" title="Hiérarchie incorrecte">!</span>
{% endif %}
</li>
{% endfor %}
@ -100,7 +121,9 @@
<li class="list-group-item list-group-item-info title-level-{{ section.level }}">
{# <span class="badge bg-secondary">h{{ section.level }}</span> #}
{{ section.title }}
{% if detailed_comparison.fr_hierarchy_errors is defined and (detailed_comparison.section_comparison.common|length + loop.index0) in detailed_comparison.fr_hierarchy_errors %}
<span class="badge bg-danger ms-2" title="Hiérarchie incorrecte">!</span>
{% endif %}
</li>
{% endfor %}
</ul>
@ -227,59 +250,41 @@
<div class="card-body">
<h4>Liens uniquement en anglais
({{ detailed_comparison.link_comparison.en_only|length }})</h4>
<h4>Comparaison des liens côte à côte</h4>
<div class="table-responsive">
<table class="table table-sm table-warning">
<table class="table table-sm">
<thead>
<tr>
<th>Texte</th>
<th>URL</th>
<th class="bg-primary text-white">Texte EN</th>
<th class="bg-primary text-white">URL EN</th>
<th class="bg-info text-white">Texte FR</th>
<th class="bg-info text-white">URL FR</th>
</tr>
</thead>
<tbody>
{% for link in detailed_comparison.link_comparison.en_only %}
{% set en_links = detailed_comparison.link_comparison.en_only %}
{% set fr_links = detailed_comparison.link_comparison.fr_only %}
{% set max_links = max(en_links|length, fr_links|length) %}
{% for i in 0..(max_links - 1) %}
<tr>
<td>{{ link.text }}</td>
<td><a href="{{ link.href }}" target="_blank"
class="small">{{ link.href|slice(0, 30) }}...</a></td>
{% if i < en_links|length %}
<td class="bg-light">{{ en_links[i].text }}</td>
<td class="bg-light"><a href="{{ en_links[i].href }}" target="_blank" class="small">{{ en_links[i].href|slice(0, 30) }}...</a></td>
{% else %}
<td class="bg-light"></td>
<td class="bg-light"></td>
{% endif %}
{% if i < fr_links|length %}
<td>{{ fr_links[i].text }}</td>
<td><a href="{{ fr_links[i].href }}" target="_blank" class="small">{{ fr_links[i].href|slice(0, 30) }}...</a></td>
{% else %}
<td></td>
<td></td>
{% endif %}
</tr>
{% endfor %}
</tbody>
</table>
</div>
</div>
</div>
</div>
<div class="col-md-6">
<div class="card h-100">
<div class="card-header bg-info text-white">
<h3>Liens en français</h3>
<span class="badge bg-light text-dark">{{ fr_page.link_count }} liens</span>
</div>
<div class="card-body">
<h4>Liens uniquement en français
({{ detailed_comparison.link_comparison.fr_only|length }})</h4>
<div class="table-responsive">
<table class="table table-sm table-info">
<thead>
<tr>
<th>Texte</th>
<th>URL</th>
</tr>
</thead>
<tbody>
{% for link in detailed_comparison.link_comparison.fr_only %}
<tr>
<td>{{ link.text }}</td>
<td><a href="{{ link.href }}" target="_blank"
class="small">{{ link.href|slice(0, 30) }}...</a></td>
</tr>
{% endfor %}
</tbody>
</table>
</div>
@ -377,7 +382,13 @@
<ul class="list-group mb-3">
<li class="list-group-item d-flex justify-content-between align-items-center">
Sections
<span class="badge bg-primary rounded-pill">{{ en_page.sections }}</span>
<span class="badge bg-primary rounded-pill">
{% if detailed_comparison.adjusted_en_section_count is defined %}
{{ detailed_comparison.adjusted_en_section_count }}
{% else %}
{{ en_page.sections }}
{% endif %}
</span>
</li>
<li class="list-group-item d-flex justify-content-between align-items-center">
Mots
@ -510,7 +521,13 @@
<ul class="list-group mb-3">
<li class="list-group-item d-flex justify-content-between align-items-center">
Sections
<span class="badge bg-primary rounded-pill">{{ en_page.sections }}</span>
<span class="badge bg-primary rounded-pill">
{% if detailed_comparison.adjusted_en_section_count is defined %}
{{ detailed_comparison.adjusted_en_section_count }}
{% else %}
{{ en_page.sections }}
{% endif %}
</span>
</li>
<li class="list-group-item d-flex justify-content-between align-items-center">
Mots
@ -547,7 +564,13 @@
<ul class="list-group mb-3">
<li class="list-group-item d-flex justify-content-between align-items-center">
Sections
<span class="badge bg-info rounded-pill">{{ fr_page.sections }}</span>
<span class="badge bg-info rounded-pill">
{% if detailed_comparison.adjusted_fr_section_count is defined %}
{{ detailed_comparison.adjusted_fr_section_count }}
{% else %}
{{ fr_page.sections }}
{% endif %}
</span>
</li>
<li class="list-group-item d-flex justify-content-between align-items-center">
Mots

View file

@ -96,26 +96,103 @@
</div>
<div class="card-body">
{% if local_groups|length > 0 %}
<!-- Filtres -->
<div class="mb-4">
<div class="btn-group" role="group" aria-label="Filtres">
<button type="button" class="btn btn-outline-primary active filter-btn" data-filter="all">Tous</button>
<button type="button" class="btn btn-outline-primary filter-btn" data-filter="wiki">Wiki</button>
<button type="button" class="btn btn-outline-primary filter-btn" data-filter="framacalc">Framacalc</button>
<button type="button" class="btn btn-outline-primary filter-btn" data-filter="has-wiki">Avec page wiki</button>
<button type="button" class="btn btn-outline-primary filter-btn" data-filter="no-wiki">Sans page wiki</button>
</div>
</div>
<div class="row row-cols-1 row-cols-md-2 row-cols-lg-3 g-4">
{% for group in local_groups %}
<div class="col">
<div class="card h-100">
{% set source = group.source|default('wiki') %}
{% set has_wiki = group.has_wiki_page|default(true) %}
{% set filter_classes = source ~ ' ' ~ (has_wiki ? 'has-wiki' : 'no-wiki') %}
<div class="col group-item {{ filter_classes }}">
<div class="card h-100 {% if source == 'framacalc' and not has_wiki %}border-danger{% endif %}">
{% if source == 'framacalc' %}
<div class="card-header bg-light">
<span class="badge bg-secondary">Framacalc</span>
{% if has_wiki %}
<span class="badge bg-success">Page wiki</span>
{% else %}
<span class="badge bg-danger">Pas de page wiki</span>
{% endif %}
</div>
{% endif %}
<div class="card-body">
<h5 class="card-title">{{ group.name }}</h5>
{% if group.description %}
<p class="card-text">{{ group.description }}</p>
{% endif %}
{% if source == 'framacalc' and group.contact %}
<p class="card-text"><small class="text-muted">Contact: {{ group.contact }}</small></p>
{% endif %}
{% if source == 'framacalc' and group.website %}
<p class="card-text">
<a href="{{ group.website }}" target="_blank" class="btn btn-sm btn-outline-secondary">
<i class="bi bi-globe"></i> Site web
</a>
</p>
{% endif %}
</div>
<div class="card-footer">
<a href="{{ group.url }}" target="_blank"
class="btn btn-sm btn-outline-primary">
<i class="bi bi-box-arrow-up-right"></i> Voir sur le wiki
</a>
{% if source == 'wiki' or has_wiki %}
<a href="{{ group.url }}" target="_blank"
class="btn btn-sm btn-outline-primary">
<i class="bi bi-box-arrow-up-right"></i> Voir sur le wiki
</a>
{% else %}
<a href="https://wiki.openstreetmap.org/wiki/Special:Search?search={{ group.name|url_encode }}"
target="_blank" class="btn btn-sm btn-outline-danger">
<i class="bi bi-search"></i> Rechercher sur le wiki
</a>
{% endif %}
</div>
</div>
</div>
{% endfor %}
</div>
<!-- JavaScript pour les filtres -->
<script>
document.addEventListener('DOMContentLoaded', function() {
const filterButtons = document.querySelectorAll('.filter-btn');
const groupItems = document.querySelectorAll('.group-item');
filterButtons.forEach(button => {
button.addEventListener('click', function() {
// Remove active class from all buttons
filterButtons.forEach(btn => btn.classList.remove('active'));
// Add active class to clicked button
this.classList.add('active');
const filter = this.getAttribute('data-filter');
// Show/hide items based on filter
groupItems.forEach(item => {
if (filter === 'all') {
item.style.display = 'block';
} else {
if (item.classList.contains(filter)) {
item.style.display = 'block';
} else {
item.style.display = 'none';
}
}
});
});
});
});
</script>
{% else %}
<div class="alert alert-info">
<p><i class="bi bi-info-circle"></i> Aucun groupe local n'a été trouvé.</p>

View file

@ -3,120 +3,122 @@
{% block title %}Suggestion de page Wiki à améliorer{% endblock %}
{% block body %}
<div class="container mt-4">
{% include 'admin/_wiki_navigation.html.twig' %}
<h1>Suggestion de page Wiki à améliorer</h1>
<p class="lead">Voici une page wiki qui a besoin d'être améliorée.</p>
<div class="container mt-4">
{% include 'admin/_wiki_navigation.html.twig' %}
<div class="card mb-4">
<div class="card-header bg-primary text-white">
<h2>{{ page.key }}</h2>
</div>
<div class="card-body">
<div class="alert alert-info">
<h3>Raisons d'amélioration</h3>
<p>{{ page.reason }}</p>
<h1>Suggestion de page Wiki à améliorer</h1>
<p class="lead">Voici une page wiki qui a besoin d'être améliorée.</p>
<div class="card mb-4">
<div class="card-header bg-primary text-white">
<h2>{{ page.key }}</h2>
</div>
<div class="card-body">
<div class="alert alert-info">
<h3>Raisons d'amélioration</h3>
<p>{{ page.reason }}</p>
</div>
<div class="row">
<div class="col-md-6">
<div class="card h-100">
<div class="card-header bg-primary text-white">
<h3>Version anglaise</h3>
<p class="mb-0">
<small>Dernière modification: {{ page.en_page.last_modified }}</small>
</p>
<div class="row">
<div class="col-md-6">
<div class="card h-100">
<div class="card-header bg-primary text-white">
<h3>Version anglaise</h3>
<p class="mb-0">
<small>Dernière modification: {{ page.en_page.last_modified }}</small>
</p>
</div>
<div class="card-body">
<ul class="list-group mb-3">
<li class="list-group-item d-flex justify-content-between align-items-center">
Sections
<span class="badge bg-primary rounded-pill">{{ page.en_page.sections }}</span>
</li>
<li class="list-group-item d-flex justify-content-between align-items-center">
Mots
<span class="badge bg-primary rounded-pill">{{ page.en_page.word_count|default(0) }}</span>
</li>
<li class="list-group-item d-flex justify-content-between align-items-center">
Liens
<span class="badge bg-primary rounded-pill">{{ page.en_page.link_count|default(0) }}</span>
</li>
</ul>
<div class="d-grid gap-2">
<a href="{{ page.en_page.url }}" target="_blank" class="btn btn-outline-primary">
<i class="bi bi-box-arrow-up-right"></i> Voir la page anglaise
</a>
</div>
</div>
</div>
<div class="card-body">
<ul class="list-group mb-3">
<li class="list-group-item d-flex justify-content-between align-items-center">
Sections
<span class="badge bg-primary rounded-pill">{{ page.en_page.sections }}</span>
</li>
<li class="list-group-item d-flex justify-content-between align-items-center">
Mots
<span class="badge bg-primary rounded-pill">{{ page.en_page.word_count|default(0) }}</span>
</li>
<li class="list-group-item d-flex justify-content-between align-items-center">
Liens
<span class="badge bg-primary rounded-pill">{{ page.en_page.link_count|default(0) }}</span>
</li>
</ul>
<div class="d-grid gap-2">
<a href="{{ page.en_page.url }}" target="_blank" class="btn btn-outline-primary">
<i class="bi bi-box-arrow-up-right"></i> Voir la page anglaise
</a>
</div>
<div class="col-md-6">
<div class="card h-100">
<div class="card-header bg-info text-white">
<h3>Version française</h3>
{% if page.fr_page %}
<p class="mb-0">
<small>Dernière modification: {{ page.fr_page.last_modified }}</small>
</p>
{% else %}
<p class="mb-0">
<small>Page non existante</small>
</p>
{% endif %}
</div>
<div class="card-body">
{% if page.fr_page %}
<ul class="list-group mb-3">
<li class="list-group-item d-flex justify-content-between align-items-center">
Sections
<span class="badge bg-info rounded-pill">{{ page.fr_page.sections }}</span>
</li>
<li class="list-group-item d-flex justify-content-between align-items-center">
Mots
<span class="badge bg-info rounded-pill">{{ page.fr_page.word_count|default(0) }}</span>
</li>
<li class="list-group-item d-flex justify-content-between align-items-center">
Liens
<span class="badge bg-info rounded-pill">{{ page.fr_page.link_count|default(0) }}</span>
</li>
</ul>
<div class="d-grid gap-2">
<a href="{{ page.fr_page.url }}" target="_blank" class="btn btn-outline-info">
<i class="bi bi-box-arrow-up-right"></i> Voir la page française
</a>
</div>
{% else %}
<div class="alert alert-warning">
<p><i class="bi bi-exclamation-triangle"></i> <strong>La page wiki pour la clé
"{{ page.key }}" n'existe pas en français.</strong></p>
<p>Vous pouvez contribuer en créant cette page sur le wiki OpenStreetMap.</p>
</div>
<div class="d-grid gap-2">
<a href="https://wiki.openstreetmap.org/w/index.php?title=FR:Key:{{ page.key }}&action=edit"
target="_blank" class="btn btn-success">
<i class="bi bi-plus-circle"></i> Créer la page française
</a>
</div>
{% endif %}
</div>
</div>
</div>
</div>
<div class="col-md-6">
<div class="card h-100">
<div class="card-header bg-info text-white">
<h3>Version française</h3>
{% if page.fr_page %}
<p class="mb-0">
<small>Dernière modification: {{ page.fr_page.last_modified }}</small>
</p>
{% else %}
<p class="mb-0">
<small>Page non existante</small>
</p>
{% endif %}
</div>
<div class="card-body">
{% if page.fr_page %}
<ul class="list-group mb-3">
<li class="list-group-item d-flex justify-content-between align-items-center">
Sections
<span class="badge bg-info rounded-pill">{{ page.fr_page.sections }}</span>
</li>
<li class="list-group-item d-flex justify-content-between align-items-center">
Mots
<span class="badge bg-info rounded-pill">{{ page.fr_page.word_count|default(0) }}</span>
</li>
<li class="list-group-item d-flex justify-content-between align-items-center">
Liens
<span class="badge bg-info rounded-pill">{{ page.fr_page.link_count|default(0) }}</span>
</li>
</ul>
<div class="d-grid gap-2">
<a href="{{ page.fr_page.url }}" target="_blank" class="btn btn-outline-info">
<i class="bi bi-box-arrow-up-right"></i> Voir la page française
</a>
</div>
{% else %}
<div class="alert alert-warning">
<p><i class="bi bi-exclamation-triangle"></i> <strong>La page wiki pour la clé "{{ page.key }}" n'existe pas en français.</strong></p>
<p>Vous pouvez contribuer en créant cette page sur le wiki OpenStreetMap.</p>
</div>
<div class="d-grid gap-2">
<a href="https://wiki.openstreetmap.org/wiki/FR:Key:{{ page.key }}" target="_blank" class="btn btn-success">
<i class="bi bi-plus-circle"></i> Créer la page française
</a>
</div>
{% endif %}
</div>
</div>
<div class="mt-4 d-grid gap-2">
<a href="{{ path('app_admin_wiki_compare', {'key': page.key}) }}" class="btn btn-primary">
<i class="bi bi-arrows-angle-expand"></i> Voir la comparaison détaillée
</a>
<a href="{{ path('app_admin_wiki_random_suggestion') }}" class="btn btn-secondary">
<i class="bi bi-shuffle"></i> Autre suggestion aléatoire
</a>
</div>
</div>
</div>
<div class="mt-4 d-grid gap-2">
<a href="{{ path('app_admin_wiki_compare', {'key': page.key}) }}" class="btn btn-primary">
<i class="bi bi-arrows-angle-expand"></i> Voir la comparaison détaillée
</a>
<a href="{{ path('app_admin_wiki_random_suggestion') }}" class="btn btn-secondary">
<i class="bi bi-shuffle"></i> Autre suggestion aléatoire
</a>
</div>
<div class="mt-3">
<a href="{{ path('app_admin_wiki') }}" class="btn btn-secondary">
<i class="bi bi-arrow-left"></i> Retour à la liste des pages wiki
</a>
</div>
</div>
<div class="mt-3">
<a href="{{ path('app_admin_wiki') }}" class="btn btn-secondary">
<i class="bi bi-arrow-left"></i> Retour à la liste des pages wiki
</a>
</div>
</div>
{% endblock %}

View file

@ -0,0 +1,87 @@
{% extends 'base.html.twig' %}
{% block title %}Changements récents Wiki OSM{% endblock %}
{% block body %}
<div class="container mt-4">
{% include 'admin/_wiki_navigation.html.twig' %}
<h1>Changements récents Wiki OpenStreetMap</h1>
<p class="lead">Liste des changements récents dans l'espace de noms français du wiki OpenStreetMap.</p>
{% if last_updated %}
<div class="alert alert-info">
<i class="bi bi-info-circle"></i> Dernière mise à jour : {{ last_updated|date('d/m/Y H:i') }}
</div>
{% endif %}
<div class="card mb-4">
<div class="card-header bg-primary text-white">
<h2>Changements récents</h2>
<p class="mb-0">
<a href="https://wiki.openstreetmap.org/w/index.php?hidebots=1&hidepreviousrevisions=1&hidecategorization=1&hideWikibase=1&hidelog=1&hidenewuserlog=1&namespace=202&limit=500&days=30&enhanced=1&title=Special:RecentChanges&urlversion=2"
target="_blank" class="text-white">
<i class="bi bi-box-arrow-up-right"></i> Voir sur le wiki OSM
</a>
</p>
</div>
<div class="card-body">
{% if recent_changes|length > 0 %}
<div class="table-responsive">
<table class="table table-striped table-hover">
<thead class="thead-dark">
<tr>
<th>Page</th>
<th>Date</th>
<th>Utilisateur</th>
<th>Commentaire</th>
<th>Taille</th>
<th>Actions</th>
</tr>
</thead>
<tbody>
{% for change in recent_changes %}
<tr>
<td>
<strong>{{ change.page_name }}</strong>
</td>
<td>{{ change.timestamp }}</td>
<td>{{ change.user }}</td>
<td>{{ change.comment }}</td>
<td>{{ change.change_size }}</td>
<td>
<a href="{{ change.page_url }}" target="_blank" class="btn btn-sm btn-outline-primary">
<i class="bi bi-box-arrow-up-right"></i> Voir
</a>
</td>
</tr>
{% endfor %}
</tbody>
</table>
</div>
{% else %}
<div class="alert alert-info">
<p><i class="bi bi-info-circle"></i> Aucun changement récent n'a été trouvé.</p>
</div>
{% endif %}
</div>
</div>
<div class="card mb-4">
<div class="card-header">
<h2>À propos des changements récents</h2>
</div>
<div class="card-body">
<p>Cette page affiche les changements récents dans l'espace de noms français (FR:) du wiki OpenStreetMap.</p>
<p>Ces informations sont utiles pour suivre les traductions manquantes et les mises à jour des pages wiki.</p>
<p>Les données sont mises à jour automatiquement toutes les heures.</p>
</div>
</div>
<div class="mt-3">
<a href="{{ path('app_admin_wiki') }}" class="btn btn-secondary">
<i class="bi bi-arrow-left"></i> Retour à la liste des pages wiki
</a>
</div>
</div>
{% endblock %}

66
wiki_compare/CHANGES.md Normal file
View file

@ -0,0 +1,66 @@
# Changements implémentés
Ce document résume les changements et nouvelles fonctionnalités implémentés dans le cadre de la mise à jour du système de gestion des pages wiki OSM.
## 1. Suivi des changements récents du wiki OSM
### Fonctionnalités ajoutées
- Création d'un script `fetch_recent_changes.py` qui récupère les changements récents dans l'espace de noms français du wiki OSM
- Ajout d'une nouvelle route `/wiki/recent-changes` dans le contrôleur WikiController
- Création d'un template `wiki_recent_changes.html.twig` pour afficher les changements récents
- Mise à jour de la navigation pour inclure un lien vers la page des changements récents
### Utilisation
- Les changements récents sont automatiquement récupérés toutes les heures
- La page affiche la liste des pages modifiées récemment avec des liens vers ces pages
## 2. Validation de la hiérarchie des titres
### Fonctionnalités ajoutées
- Implémentation d'une logique de détection des hiérarchies de titres incorrectes (par exemple, h4 directement sous h2 sans h3 intermédiaire)
- Ajout d'indicateurs visuels (badges) pour signaler les hiérarchies incorrectes dans les listes de sections
- Mise à jour du template `wiki_compare.html.twig` pour afficher ces indicateurs
### Utilisation
- Les hiérarchies incorrectes sont automatiquement détectées lors de la comparaison des pages wiki
- Un badge rouge avec un point d'exclamation est affiché à côté des titres ayant une hiérarchie incorrecte
## 3. Vérification des groupes locaux
### Fonctionnalités ajoutées
- Mise à jour du script `fetch_osm_fr_groups.py` pour récupérer les données des groupes locaux depuis Framacalc
- Ajout d'une fonctionnalité de vérification de l'existence d'une page wiki pour chaque groupe
- Mise à jour du template `wiki_osm_fr_groups.html.twig` pour afficher les résultats de vérification
- Ajout de filtres pour faciliter la navigation parmi les groupes
### Utilisation
- Les groupes sont affichés avec des badges indiquant leur source (wiki ou Framacalc)
- Les groupes sans page wiki sont mis en évidence avec un badge rouge
- Les filtres permettent de voir uniquement les groupes d'une certaine catégorie (tous, wiki, Framacalc, avec page wiki, sans page wiki)
## Limitations connues
1. **Accès aux données externes** : Les scripts peuvent rencontrer des difficultés pour accéder aux données externes (wiki OSM, Framacalc) en fonction de l'environnement d'exécution.
2. **Détection des hiérarchies** : La détection des hiérarchies incorrectes se base uniquement sur les niveaux des titres et ne prend pas en compte le contenu ou la sémantique.
3. **Correspondance des groupes** : La correspondance entre les groupes Framacalc et les pages wiki se fait par une comparaison approximative des noms, ce qui peut parfois donner des résultats imprécis.
## Maintenance future
### Scripts Python
- Les scripts Python sont situés dans le répertoire `wiki_compare/`
- Ils peuvent être exécutés manuellement ou via des tâches cron
- L'option `--dry-run` permet de tester les scripts sans modifier les fichiers
- L'option `--force` permet de forcer la mise à jour même si le cache est récent
### Templates Twig
- Les templates sont situés dans le répertoire `templates/admin/`
- `wiki_recent_changes.html.twig` : Affichage des changements récents
- `wiki_compare.html.twig` : Comparaison des pages wiki avec validation de hiérarchie
- `wiki_osm_fr_groups.html.twig` : Affichage des groupes locaux avec vérification des pages wiki
### Contrôleur
- Le contrôleur `WikiController.php` contient toutes les routes et la logique de traitement
- La méthode `detectHeadingHierarchyErrors()` peut être ajustée pour modifier les règles de validation des hiérarchies
- Les méthodes de rafraîchissement des données (`refreshRecentChangesData()`, etc.) peuvent être modifiées pour ajuster la fréquence de mise à jour

View file

@ -4,9 +4,11 @@
"""
fetch_osm_fr_groups.py
This script scrapes the OpenStreetMap wiki page for France/OSM-FR to extract
information about local working groups. It specifically targets links in the
#Pages_des_groupes_locaux section.
This script fetches information about OSM-FR local groups from two sources:
1. The OpenStreetMap wiki page for France/OSM-FR (specifically the #Pages_des_groupes_locaux section)
2. The Framacalc spreadsheet at https://framacalc.org/osm-groupes-locaux
It then verifies that each group from the Framacalc has a corresponding wiki page.
Usage:
python fetch_osm_fr_groups.py [--dry-run] [--force]
@ -24,6 +26,8 @@ import json
import argparse
import logging
import os
import csv
import io
from datetime import datetime, timedelta
import requests
from bs4 import BeautifulSoup
@ -40,6 +44,8 @@ logger = logging.getLogger(__name__)
OUTPUT_FILE = "osm_fr_groups.json"
BASE_URL = "https://wiki.openstreetmap.org/wiki/France/OSM-FR"
WIKI_BASE_URL = "https://wiki.openstreetmap.org"
FRAMACALC_URL = "https://framacalc.org/osm-groupes-locaux/export/csv"
WIKI_GROUPS_URL = "https://wiki.openstreetmap.org/wiki/France/OSM-FR#Groupes_locaux"
CACHE_DURATION = timedelta(hours=1) # Cache duration of 1 hour
def is_cache_fresh():
@ -141,7 +147,7 @@ def extract_working_groups(html_content):
logger.info(f"Found {len(working_groups)} working groups")
return working_groups
def extract_local_groups(html_content):
def extract_local_groups_from_wiki(html_content):
"""
Extract local groups from the wiki page HTML
@ -193,13 +199,170 @@ def extract_local_groups(html_content):
"name": name,
"url": url,
"description": description,
"type": "local_group"
"type": "local_group",
"source": "wiki"
})
current = current.next_sibling
logger.info(f"Found {len(local_groups)} local groups")
logger.info(f"Found {len(local_groups)} local groups from wiki")
return local_groups
def fetch_framacalc_data():
"""
Fetch local groups data from Framacalc
Returns:
list: List of local group dictionaries from Framacalc
"""
try:
response = requests.get(FRAMACALC_URL)
response.raise_for_status()
# Parse CSV data
csv_data = csv.reader(io.StringIO(response.text))
rows = list(csv_data)
# Check if we have data
if len(rows) < 2:
logger.warning("No data found in Framacalc CSV")
return []
# Extract headers (first row)
headers = rows[0]
# Find the indices of important columns
name_idx = -1
contact_idx = -1
website_idx = -1
for i, header in enumerate(headers):
header_lower = header.lower()
if 'nom' in header_lower or 'groupe' in header_lower:
name_idx = i
elif 'contact' in header_lower or 'email' in header_lower:
contact_idx = i
elif 'site' in header_lower or 'web' in header_lower:
website_idx = i
if name_idx == -1:
logger.warning("Could not find name column in Framacalc CSV")
return []
# Process data rows
local_groups = []
for row in rows[1:]: # Skip header row
if len(row) <= name_idx or not row[name_idx].strip():
continue # Skip empty rows
name = row[name_idx].strip()
contact = row[contact_idx].strip() if contact_idx != -1 and contact_idx < len(row) else ""
website = row[website_idx].strip() if website_idx != -1 and website_idx < len(row) else ""
local_groups.append({
"name": name,
"contact": contact,
"website": website,
"type": "local_group",
"source": "framacalc",
"has_wiki_page": False, # Will be updated later
"wiki_url": "" # Will be updated later
})
logger.info(f"Found {len(local_groups)} local groups from Framacalc")
return local_groups
except requests.exceptions.RequestException as e:
logger.error(f"Error fetching Framacalc data: {e}")
return []
except Exception as e:
logger.error(f"Error processing Framacalc data: {e}")
return []
def extract_wiki_group_links():
"""
Extract links to local group wiki pages from the OSM-FR wiki page
Returns:
dict: Dictionary mapping group names to wiki URLs
"""
try:
# Get the wiki page content
response = requests.get(WIKI_GROUPS_URL)
response.raise_for_status()
soup = BeautifulSoup(response.text, 'html.parser')
wiki_links = {}
# Find the "Pages des groupes locaux" section
pages_section = None
for heading in soup.find_all(['h2', 'h3', 'h4']):
if 'Pages des groupes locaux' in heading.get_text():
pages_section = heading
break
if not pages_section:
logger.warning("Could not find 'Pages des groupes locaux' section")
return {}
# Get the content following the heading until the next heading
current = pages_section.next_sibling
while current and not current.name in ['h2', 'h3', 'h4']:
if current.name == 'ul':
# Process list items
for li in current.find_all('li', recursive=False):
text = li.get_text().strip()
link = li.find('a')
if link and text:
# Extract group name (before the comma)
parts = text.split(',', 1)
group_name = parts[0].strip()
url = WIKI_BASE_URL + link.get('href') if link.get('href').startswith('/') else link.get('href')
wiki_links[group_name] = url
current = current.next_sibling
logger.info(f"Found {len(wiki_links)} wiki links for local groups")
return wiki_links
except requests.exceptions.RequestException as e:
logger.error(f"Error fetching wiki group links: {e}")
return {}
except Exception as e:
logger.error(f"Error processing wiki group links: {e}")
return {}
def verify_framacalc_groups_have_wiki(framacalc_groups, wiki_links):
"""
Verify that each group from Framacalc has a corresponding wiki page
Args:
framacalc_groups (list): List of local group dictionaries from Framacalc
wiki_links (dict): Dictionary mapping group names to wiki URLs
Returns:
list: Updated list of local group dictionaries with wiki verification
"""
for group in framacalc_groups:
group_name = group['name']
# Try to find a matching wiki link
found = False
for wiki_name, wiki_url in wiki_links.items():
# Check if the group name is similar to the wiki name
if group_name.lower() in wiki_name.lower() or wiki_name.lower() in group_name.lower():
group['has_wiki_page'] = True
group['wiki_url'] = wiki_url
found = True
break
if not found:
group['has_wiki_page'] = False
group['wiki_url'] = ""
return framacalc_groups
def extract_umap_url(html_content):
"""
Extract the uMap URL for OSM-FR local groups
@ -223,14 +386,16 @@ def extract_umap_url(html_content):
return None
def save_results(local_groups, working_groups, umap_url, dry_run=False):
def save_results(wiki_local_groups, framacalc_groups, working_groups, umap_url, wiki_links, dry_run=False):
"""
Save the results to a JSON file
Args:
local_groups (list): List of local group dictionaries
wiki_local_groups (list): List of local group dictionaries from wiki
framacalc_groups (list): List of local group dictionaries from Framacalc
working_groups (list): List of working group dictionaries
umap_url (str): URL to the uMap for local groups
wiki_links (dict): Dictionary mapping group names to wiki URLs
dry_run (bool): If True, don't actually save to file
Returns:
@ -238,28 +403,41 @@ def save_results(local_groups, working_groups, umap_url, dry_run=False):
"""
if dry_run:
logger.info("DRY RUN: Would have saved results to file")
logger.info(f"Local groups: {len(local_groups)}")
for group in local_groups:
logger.info(f"Wiki local groups: {len(wiki_local_groups)}")
for group in wiki_local_groups[:5]: # Show only first 5 for brevity
logger.info(f" - {group['name']}: {group['url']}")
logger.info(f"Framacalc groups: {len(framacalc_groups)}")
for group in framacalc_groups[:5]: # Show only first 5 for brevity
wiki_status = "Has wiki page" if group.get('has_wiki_page') else "No wiki page"
logger.info(f" - {group['name']}: {wiki_status}")
logger.info(f"Working groups: {len(working_groups)}")
for group in working_groups:
for group in working_groups[:5]: # Show only first 5 for brevity
logger.info(f" - {group['name']}: {group['url']}")
if umap_url:
logger.info(f"uMap URL: {umap_url}")
logger.info(f"Wiki links: {len(wiki_links)}")
return True
# Combine all local groups
all_local_groups = wiki_local_groups + framacalc_groups
# Prepare the data structure
data = {
"last_updated": datetime.now().isoformat(),
"local_groups": local_groups,
"local_groups": all_local_groups,
"working_groups": working_groups,
"umap_url": umap_url
"umap_url": umap_url,
"wiki_links": wiki_links
}
try:
with open(OUTPUT_FILE, 'w', encoding='utf-8') as f:
json.dump(data, f, indent=2, ensure_ascii=False)
logger.info(f"Successfully saved {len(local_groups)} local groups and {len(working_groups)} working groups to {OUTPUT_FILE}")
logger.info(f"Successfully saved {len(all_local_groups)} local groups and {len(working_groups)} working groups to {OUTPUT_FILE}")
return True
except IOError as e:
logger.error(f"Error saving results to {OUTPUT_FILE}: {e}")
@ -267,7 +445,7 @@ def save_results(local_groups, working_groups, umap_url, dry_run=False):
def main():
"""Main function to execute the script"""
parser = argparse.ArgumentParser(description="Scrape OSM-FR local groups from the wiki")
parser = argparse.ArgumentParser(description="Fetch OSM-FR local groups from wiki and Framacalc")
parser.add_argument("--dry-run", action="store_true", help="Run without saving results to file")
parser.add_argument("--force", action="store_true", help="Force update even if cache is fresh")
args = parser.parse_args()
@ -287,11 +465,11 @@ def main():
logger.error("Failed to get wiki page content")
return
# Extract local groups
local_groups = extract_local_groups(html_content)
# Extract local groups from wiki
wiki_local_groups = extract_local_groups_from_wiki(html_content)
if not local_groups:
logger.warning("No local groups found")
if not wiki_local_groups:
logger.warning("No local groups found in wiki")
# Extract working groups
working_groups = extract_working_groups(html_content)
@ -304,8 +482,31 @@ def main():
# Extract uMap URL
umap_url = extract_umap_url(html_content)
# Fetch local groups from Framacalc
framacalc_groups = fetch_framacalc_data()
if not framacalc_groups:
logger.warning("No local groups found in Framacalc")
# Extract wiki group links
wiki_links = extract_wiki_group_links()
if not wiki_links:
logger.warning("No wiki links found for local groups")
# Verify Framacalc groups have wiki pages
if framacalc_groups and wiki_links:
framacalc_groups = verify_framacalc_groups_have_wiki(framacalc_groups, wiki_links)
# Count groups with and without wiki pages
groups_with_wiki = sum(1 for group in framacalc_groups if group.get('has_wiki_page'))
groups_without_wiki = sum(1 for group in framacalc_groups if not group.get('has_wiki_page'))
logger.info(f"Framacalc groups with wiki pages: {groups_with_wiki}")
logger.info(f"Framacalc groups without wiki pages: {groups_without_wiki}")
# Save results
success = save_results(local_groups, working_groups, umap_url, args.dry_run)
success = save_results(wiki_local_groups, framacalc_groups, working_groups, umap_url, wiki_links, args.dry_run)
if success:
logger.info("Script completed successfully")

View file

@ -0,0 +1,216 @@
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
fetch_recent_changes.py
This script fetches recent changes from the OpenStreetMap wiki for the French namespace
and stores the URLs of these pages. It specifically targets the recent changes page:
https://wiki.openstreetmap.org/w/index.php?hidebots=1&hidepreviousrevisions=1&hidecategorization=1&hideWikibase=1&hidelog=1&hidenewuserlog=1&namespace=202&limit=500&days=30&enhanced=1&title=Special:RecentChanges&urlversion=2
Usage:
python fetch_recent_changes.py [--dry-run] [--force]
Options:
--dry-run Run the script without saving the results to a file
--force Force update even if the cache is still fresh (less than 1 hour old)
Output:
- recent_changes.json: JSON file with information about recent changes in the French namespace
- Log messages about the scraping process and results
"""
import json
import argparse
import logging
import os
from datetime import datetime, timedelta
import requests
from bs4 import BeautifulSoup
# Configure logging
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(levelname)s - %(message)s',
datefmt='%Y-%m-%d %H:%M:%S'
)
logger = logging.getLogger(__name__)
# Constants
OUTPUT_FILE = "recent_changes.json"
RECENT_CHANGES_URL = "https://wiki.openstreetmap.org/w/index.php?hidebots=1&hidepreviousrevisions=1&hidecategorization=1&hideWikibase=1&hidelog=1&hidenewuserlog=1&namespace=202&limit=500&days=30&enhanced=1&title=Special:RecentChanges&urlversion=2"
WIKI_BASE_URL = "https://wiki.openstreetmap.org"
CACHE_DURATION = timedelta(hours=1) # Cache duration of 1 hour
def is_cache_fresh():
"""
Check if the cache file exists and is less than CACHE_DURATION old
Returns:
bool: True if cache is fresh, False otherwise
"""
if not os.path.exists(OUTPUT_FILE):
return False
try:
with open(OUTPUT_FILE, 'r', encoding='utf-8') as f:
data = json.load(f)
last_updated = datetime.fromisoformat(data.get('last_updated', '2000-01-01T00:00:00'))
now = datetime.now()
return (now - last_updated) < CACHE_DURATION
except (IOError, json.JSONDecodeError, ValueError) as e:
logger.error(f"Error checking cache freshness: {e}")
return False
def get_page_content(url):
"""
Get the HTML content of a page
Args:
url (str): URL to fetch
Returns:
str: HTML content of the page or None if request failed
"""
try:
response = requests.get(url)
response.raise_for_status()
return response.text
except requests.exceptions.RequestException as e:
logger.error(f"Error fetching {url}: {e}")
return None
def extract_recent_changes(html_content):
"""
Extract recent changes from the wiki page HTML
Args:
html_content (str): HTML content of the recent changes page
Returns:
list: List of recent change dictionaries
"""
if not html_content:
return []
soup = BeautifulSoup(html_content, 'html.parser')
recent_changes = []
# Find the changes list
changes_list = soup.find('ul', class_='special')
if not changes_list:
logger.warning("Could not find recent changes list")
return []
# Process each list item (each change)
for li in changes_list.find_all('li'):
# Extract the page link
page_link = li.find('a', class_='mw-changeslist-title')
if not page_link:
continue
page_name = page_link.get_text().strip()
page_url = WIKI_BASE_URL + page_link.get('href')
# Extract the timestamp
timestamp_span = li.find('span', class_='mw-changeslist-date')
timestamp = timestamp_span.get_text().strip() if timestamp_span else "Unknown"
# Extract the user
user_link = li.find('a', class_='mw-userlink')
user = user_link.get_text().strip() if user_link else "Unknown"
# Extract the comment
comment_span = li.find('span', class_='comment')
comment = comment_span.get_text().strip() if comment_span else ""
# Extract the change size
change_size_span = li.find('span', class_='mw-changeslist-separator').next_sibling
change_size = change_size_span.get_text().strip() if change_size_span else "0"
recent_changes.append({
"page_name": page_name,
"page_url": page_url,
"timestamp": timestamp,
"user": user,
"comment": comment,
"change_size": change_size
})
logger.info(f"Found {len(recent_changes)} recent changes")
return recent_changes
def save_results(recent_changes, dry_run=False):
"""
Save the results to a JSON file
Args:
recent_changes (list): List of recent change dictionaries
dry_run (bool): If True, don't actually save to file
Returns:
bool: True if saving was successful or dry run, False otherwise
"""
if dry_run:
logger.info("DRY RUN: Would have saved results to file")
logger.info(f"Recent changes: {len(recent_changes)}")
for change in recent_changes[:5]: # Show only first 5 for brevity
logger.info(f" - {change['page_name']}: {change['page_url']} ({change['timestamp']})")
if len(recent_changes) > 5:
logger.info(f" ... and {len(recent_changes) - 5} more")
return True
# Prepare the data structure
data = {
"last_updated": datetime.now().isoformat(),
"recent_changes": recent_changes
}
try:
with open(OUTPUT_FILE, 'w', encoding='utf-8') as f:
json.dump(data, f, indent=2, ensure_ascii=False)
logger.info(f"Successfully saved {len(recent_changes)} recent changes to {OUTPUT_FILE}")
return True
except IOError as e:
logger.error(f"Error saving results to {OUTPUT_FILE}: {e}")
return False
def main():
"""Main function to execute the script"""
parser = argparse.ArgumentParser(description="Fetch recent changes from the OSM wiki French namespace")
parser.add_argument("--dry-run", action="store_true", help="Run without saving results to file")
parser.add_argument("--force", action="store_true", help="Force update even if cache is fresh")
args = parser.parse_args()
logger.info("Starting fetch_recent_changes.py")
# Check if cache is fresh
if is_cache_fresh() and not args.force:
logger.info(f"Cache is still fresh (less than {CACHE_DURATION.total_seconds()/3600} hours old)")
logger.info(f"Use --force to update anyway")
return
# Get the recent changes page content
html_content = get_page_content(RECENT_CHANGES_URL)
if not html_content:
logger.error("Failed to get recent changes page content")
return
# Extract recent changes
recent_changes = extract_recent_changes(html_content)
if not recent_changes:
logger.warning("No recent changes found")
# Save results
success = save_results(recent_changes, args.dry_run)
if success:
logger.info("Script completed successfully")
else:
logger.error("Script completed with errors")
if __name__ == "__main__":
main()