fix cache dir
This commit is contained in:
parent
be66cf7e53
commit
ea0c5c7765
@ -6,7 +6,7 @@ const path = require('path');
|
||||
const BASE_URL = 'https://www.filmspourenfants.net';
|
||||
|
||||
// Setup disk cache
|
||||
const CACHE_DIR = path.join(__dirname, '../cache');
|
||||
const CACHE_DIR = process.env.CACHE_DIR || '/app/data/cache';
|
||||
if (!fs.existsSync(CACHE_DIR)) fs.mkdirSync(CACHE_DIR, { recursive: true });
|
||||
|
||||
// Cache operations
|
||||
@ -38,7 +38,7 @@ function saveCache(type, key, data) {
|
||||
// Extract age from string like "À partir de 8 ans" or "Déconseillé aux moins de: 8 ans"
|
||||
function extractAgeFromText(text) {
|
||||
if (!text) return null;
|
||||
|
||||
|
||||
const match = text.match(/(\d+)\s*ans/i);
|
||||
if (match && match[1]) {
|
||||
return parseInt(match[1]);
|
||||
@ -50,24 +50,24 @@ async function searchMovies(query) {
|
||||
// Check cache first
|
||||
const cached = loadCache('search', query);
|
||||
if (cached) return cached;
|
||||
|
||||
|
||||
const searchUrl = `${BASE_URL}/films-resultats/?_s=${encodeURIComponent(query)}`;
|
||||
console.log('Searching FilmsPourEnfants:', searchUrl);
|
||||
|
||||
|
||||
try {
|
||||
const response = await axios.get(searchUrl, {
|
||||
headers: { 'User-Agent': 'Mozilla/5.0' }
|
||||
const response = await axios.get(searchUrl, {
|
||||
headers: { 'User-Agent': 'Mozilla/5.0' }
|
||||
});
|
||||
const $ = cheerio.load(response.data);
|
||||
const results = [];
|
||||
|
||||
|
||||
$('section.gp-post-item').each((_, el) => {
|
||||
const title = $(el).find('h2.gp-loop-title a').text().trim();
|
||||
const link = $(el).find('h2.gp-loop-title a').attr('href');
|
||||
const img = $(el).find('.gp-post-thumbnail img').attr('src');
|
||||
const ageText = $(el).find('.gp-loop-cats a').text().trim();
|
||||
const age = extractAgeFromText(ageText);
|
||||
|
||||
|
||||
if (title && link) {
|
||||
results.push({
|
||||
title,
|
||||
@ -78,7 +78,7 @@ async function searchMovies(query) {
|
||||
});
|
||||
}
|
||||
});
|
||||
|
||||
|
||||
console.log(`FilmsPourEnfants found ${results.length} results`);
|
||||
saveCache('search', query, results);
|
||||
return results;
|
||||
@ -90,27 +90,27 @@ async function searchMovies(query) {
|
||||
|
||||
async function getMovieDetails(movieUrl) {
|
||||
if (!movieUrl) return {};
|
||||
|
||||
|
||||
// Check cache first
|
||||
const cached = loadCache('detail', movieUrl);
|
||||
if (cached) return cached;
|
||||
|
||||
|
||||
console.log('Fetching details for:', movieUrl);
|
||||
|
||||
|
||||
try {
|
||||
const response = await axios.get(movieUrl, {
|
||||
headers: { 'User-Agent': 'Mozilla/5.0' }
|
||||
const response = await axios.get(movieUrl, {
|
||||
headers: { 'User-Agent': 'Mozilla/5.0' }
|
||||
});
|
||||
const $ = cheerio.load(response.data);
|
||||
const details = {};
|
||||
|
||||
|
||||
// Title
|
||||
details.title = $('h1.gp-entry-title').text().trim();
|
||||
|
||||
|
||||
// Get metadata
|
||||
$('.gp-entry-meta .gp-post-meta').each((_, el) => {
|
||||
const text = $(el).text().trim();
|
||||
|
||||
|
||||
if (text.includes('Année:')) {
|
||||
details.year = $(el).find('a').text().trim();
|
||||
} else if (text.includes('Déconseillé aux moins de:')) {
|
||||
@ -125,11 +125,11 @@ async function getMovieDetails(movieUrl) {
|
||||
});
|
||||
}
|
||||
});
|
||||
|
||||
|
||||
// More detailed metadata
|
||||
$('#gp-hub-details span').each((_, el) => {
|
||||
const label = $(el).find('strong').text().trim();
|
||||
|
||||
|
||||
if (label === 'Déconseillé aux moins de:') {
|
||||
details.ageText = $(el).find('a').text().trim();
|
||||
details.age = extractAgeFromText(details.ageText);
|
||||
@ -153,13 +153,13 @@ async function getMovieDetails(movieUrl) {
|
||||
});
|
||||
}
|
||||
});
|
||||
|
||||
|
||||
// Get summary - first paragraph in the entry-text
|
||||
details.summary = $('.gp-entry-text h4').first().text().trim();
|
||||
|
||||
|
||||
// Get main image
|
||||
details.img = $('.gp-post-thumbnail img').attr('src') || $('.gp-hub-header-thumbnail img').attr('src');
|
||||
|
||||
|
||||
// Get messages section
|
||||
let messages = '';
|
||||
$('.gp-entry-text h3').each((_, el) => {
|
||||
@ -170,7 +170,7 @@ async function getMovieDetails(movieUrl) {
|
||||
}
|
||||
});
|
||||
details.messages = messages.trim();
|
||||
|
||||
|
||||
// Get difficult scenes section
|
||||
const difficultScenesHeading = $('.gp-entry-text h2:contains("SCÈNES DIFFICILES")');
|
||||
let difficultScenes = '';
|
||||
@ -184,7 +184,7 @@ async function getMovieDetails(movieUrl) {
|
||||
}
|
||||
}
|
||||
details.difficultScenes = difficultScenes.trim();
|
||||
|
||||
|
||||
console.log(`Fetched details for: ${details.title}, Age: ${details.age}`);
|
||||
saveCache('detail', movieUrl, details);
|
||||
return details;
|
||||
@ -199,7 +199,7 @@ async function searchAndEnrich(query) {
|
||||
const results = await searchMovies(query);
|
||||
return await Promise.all(results.map(async movie => {
|
||||
const details = await getMovieDetails(movie.link);
|
||||
|
||||
|
||||
return {
|
||||
title: movie.title,
|
||||
year: details.year || null,
|
||||
|
@ -5,7 +5,7 @@ const path = require('path');
|
||||
const BASE_URL = 'https://www.filmstouspublics.fr';
|
||||
|
||||
// Setup disk cache
|
||||
const CACHE_DIR = path.join(__dirname, '../cache');
|
||||
const CACHE_DIR = process.env.CACHE_DIR || '/app/data/cache';
|
||||
if (!fs.existsSync(CACHE_DIR)) fs.mkdirSync(CACHE_DIR, { recursive: true });
|
||||
|
||||
// Load cache from disk if available
|
||||
@ -46,19 +46,19 @@ function calculateAverageAge(ageRatings) {
|
||||
return !isNaN(numAge) && numAge > 0; // Only include positive ages
|
||||
})
|
||||
.map(age => typeof age === 'string' ? parseInt(age) : age);
|
||||
|
||||
|
||||
if (ages.length === 0) return null;
|
||||
|
||||
|
||||
// Calculate average
|
||||
const avg = ages.reduce((sum, age) => sum + age, 0) / ages.length;
|
||||
|
||||
|
||||
// Calculate median (more useful for skewed distributions)
|
||||
const sorted = [...ages].sort((a, b) => a - b);
|
||||
const mid = Math.floor(sorted.length / 2);
|
||||
const median = sorted.length % 2 === 0
|
||||
? (sorted[mid - 1] + sorted[mid]) / 2
|
||||
const median = sorted.length % 2 === 0
|
||||
? (sorted[mid - 1] + sorted[mid]) / 2
|
||||
: sorted[mid];
|
||||
|
||||
|
||||
return { average: avg.toFixed(1), median, countries: ages.length, min: sorted[0], max: sorted[sorted.length-1] };
|
||||
}
|
||||
|
||||
@ -66,33 +66,33 @@ async function searchMovies(query) {
|
||||
// Check cache first
|
||||
const cached = loadCache('search', query);
|
||||
if (cached) return cached;
|
||||
|
||||
|
||||
const searchUrl = `${BASE_URL}/?s=${encodeURIComponent(query)}`;
|
||||
console.log('Searching FilmsTousPublics:', searchUrl);
|
||||
|
||||
|
||||
try {
|
||||
const response = await axios.get(searchUrl, {
|
||||
headers: { 'User-Agent': 'Mozilla/5.0' }
|
||||
const response = await axios.get(searchUrl, {
|
||||
headers: { 'User-Agent': 'Mozilla/5.0' }
|
||||
});
|
||||
const $ = cheerio.load(response.data);
|
||||
const results = [];
|
||||
|
||||
|
||||
// Better selector for different article structures
|
||||
$('article[class*="tipi-xs-12"], article[class*="post"]').each((_, el) => {
|
||||
const title = $(el).find('h3.title a, .title-wrap .title a').text().trim();
|
||||
const link = $(el).find('h3.title a, .title-wrap .title a').attr('href');
|
||||
|
||||
|
||||
// Handle lazy-loaded images properly
|
||||
const imgEl = $(el).find('.mask img');
|
||||
const img = imgEl.attr('data-lazy-src') || imgEl.attr('src');
|
||||
|
||||
|
||||
// Get rating if available
|
||||
let rating = null;
|
||||
const ratingEl = $(el).find('.lets-review-api-wrap, .lets-review-final-score');
|
||||
if (ratingEl.length) {
|
||||
rating = ratingEl.attr('data-api-score') || ratingEl.text().trim();
|
||||
}
|
||||
|
||||
|
||||
if (title && link) {
|
||||
results.push({
|
||||
title,
|
||||
@ -102,7 +102,7 @@ async function searchMovies(query) {
|
||||
});
|
||||
}
|
||||
});
|
||||
|
||||
|
||||
console.log(`FilmsTousPublics found ${results.length} results`);
|
||||
saveCache('search', query, results);
|
||||
return results;
|
||||
@ -114,33 +114,33 @@ async function searchMovies(query) {
|
||||
|
||||
async function getMovieClassification(movieUrl) {
|
||||
if (!movieUrl) return {};
|
||||
|
||||
|
||||
// Check cache first
|
||||
const cached = loadCache('detail', movieUrl);
|
||||
if (cached) return cached;
|
||||
|
||||
|
||||
console.log('Fetching details for:', movieUrl);
|
||||
|
||||
|
||||
try {
|
||||
const response = await axios.get(movieUrl, {
|
||||
headers: { 'User-Agent': 'Mozilla/5.0' }
|
||||
const response = await axios.get(movieUrl, {
|
||||
headers: { 'User-Agent': 'Mozilla/5.0' }
|
||||
});
|
||||
const $ = cheerio.load(response.data);
|
||||
|
||||
|
||||
// Get country age ratings
|
||||
const ageRatings = {};
|
||||
console.log('Found pullquote elements:', $('aside.pullquote').length);
|
||||
|
||||
|
||||
// More robust approach: Find all <p> tags inside the pullquote section
|
||||
$('aside.pullquote p').each((_, el) => {
|
||||
const text = $(el).text().trim();
|
||||
console.log('Processing age text:', text);
|
||||
|
||||
|
||||
// Detect "Tous publics" for France (All audiences)
|
||||
if (text.includes('Tous publics')) {
|
||||
ageRatings.france = 0; // Set to 0 for averaging but "All" for display
|
||||
console.log('Found France rating: Tous publics (0)');
|
||||
}
|
||||
}
|
||||
// More flexible regex for the weird dashes used
|
||||
else {
|
||||
// Just extract any number that appears in string after "Déconseillé aux"
|
||||
@ -148,12 +148,12 @@ async function getMovieClassification(movieUrl) {
|
||||
if (match && match[1]) {
|
||||
const age = parseInt(match[1]);
|
||||
console.log('Found age restriction:', age);
|
||||
|
||||
|
||||
// Identify country by image alt or src
|
||||
const img = $(el).find('img');
|
||||
const alt = img.attr('alt') || '';
|
||||
const src = img.attr('src') || '';
|
||||
|
||||
|
||||
// Check for all possible countries - more flexible matching
|
||||
if (alt.includes('France') || src.toLowerCase().includes('france')) {
|
||||
ageRatings.france = age;
|
||||
@ -178,19 +178,19 @@ async function getMovieClassification(movieUrl) {
|
||||
}
|
||||
}
|
||||
});
|
||||
|
||||
|
||||
console.log('Found age ratings:', ageRatings);
|
||||
|
||||
|
||||
// Get summary/plot (first few paragraphs)
|
||||
let summary = '';
|
||||
$('.entry-content > p').each((i, el) => {
|
||||
// Skip pullquote or other non-content paragraphs
|
||||
if (!$(el).find('.pullquote').length && i < 3 && $(el).text().trim().length > 30) {
|
||||
if (!$(el).find('.pullquote').length && i < 3 && $(el).text().trim().length > 30) {
|
||||
summary += $(el).text().trim() + ' ';
|
||||
}
|
||||
});
|
||||
summary = summary.trim();
|
||||
|
||||
|
||||
// Get movie metadata
|
||||
const metadata = {};
|
||||
$('h3:contains("Informations") + ul li').each((_, el) => {
|
||||
@ -211,14 +211,14 @@ async function getMovieClassification(movieUrl) {
|
||||
metadata.studio = text.replace('Studio :', '').trim();
|
||||
}
|
||||
});
|
||||
|
||||
|
||||
// Get overall rating
|
||||
let overallRating = null;
|
||||
const ratingEl = $('.lets-review-block__final-score .score');
|
||||
if (ratingEl.length) {
|
||||
overallRating = ratingEl.text().trim();
|
||||
}
|
||||
|
||||
|
||||
// Extract year from release date if available
|
||||
let year = null;
|
||||
if (metadata.releaseDate) {
|
||||
@ -227,7 +227,7 @@ async function getMovieClassification(movieUrl) {
|
||||
year = yearMatch[0];
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
const result = {
|
||||
summary,
|
||||
year,
|
||||
@ -235,7 +235,7 @@ async function getMovieClassification(movieUrl) {
|
||||
overallRating,
|
||||
...metadata
|
||||
};
|
||||
|
||||
|
||||
// Cache the result
|
||||
saveCache('detail', movieUrl, result);
|
||||
return result;
|
||||
@ -250,11 +250,11 @@ async function searchAndEnrich(query) {
|
||||
const results = await searchMovies(query);
|
||||
return await Promise.all(results.map(async movie => {
|
||||
const details = await getMovieClassification(movie.link);
|
||||
|
||||
|
||||
// Calculate average age
|
||||
const ageStats = calculateAverageAge(details.ageRatings || {});
|
||||
console.log(`Movie: ${movie.title}, Age stats:`, ageStats);
|
||||
|
||||
|
||||
// Convert country codes to readable names for frontend display
|
||||
const countryNames = {
|
||||
france: "France",
|
||||
@ -266,22 +266,22 @@ async function searchAndEnrich(query) {
|
||||
netherlands: "Pays-Bas",
|
||||
usa: "États-Unis"
|
||||
};
|
||||
|
||||
|
||||
// Format age ratings for display
|
||||
const formattedAgeRatings = {};
|
||||
for (const [country, age] of Object.entries(details.ageRatings || {})) {
|
||||
const countryName = countryNames[country] || country;
|
||||
formattedAgeRatings[countryName] = age === 0 ? "Tous publics" : `${age}+`;
|
||||
}
|
||||
|
||||
|
||||
// Get a recommended age - prefer median, then average, then fallback
|
||||
const recommendedAge =
|
||||
const recommendedAge =
|
||||
ageStats?.median ? `${ageStats.median}+` :
|
||||
ageStats?.average ? `${ageStats.average}+` :
|
||||
details.ageRatings?.france === 0 ? "Tous publics" :
|
||||
details.ageRatings?.france ? `${details.ageRatings.france}+` :
|
||||
"Non spécifié";
|
||||
|
||||
|
||||
return {
|
||||
title: movie.title,
|
||||
year: details.year,
|
||||
@ -291,8 +291,8 @@ async function searchAndEnrich(query) {
|
||||
rating: movie.rating || details.overallRating,
|
||||
// Age information for display
|
||||
age: recommendedAge.replace('Tous publics', '0+').replace('Non spécifié', '-'),
|
||||
ageFrance: details.ageRatings?.france === 0 ? "Tous publics" :
|
||||
details.ageRatings?.france ? `${details.ageRatings.france}+` :
|
||||
ageFrance: details.ageRatings?.france === 0 ? "Tous publics" :
|
||||
details.ageRatings?.france ? `${details.ageRatings.france}+` :
|
||||
"Non spécifié",
|
||||
ageAverage: ageStats?.average || null,
|
||||
ageMedian: ageStats?.median || null,
|
||||
|
Loading…
x
Reference in New Issue
Block a user