scrap allocine to generate citebd screenings

4 files changed, 169 insertions, 0 deletions

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..85b937c
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,3 @@
+seances.xml
+cite
+cite.html
diff --git a/build.sh b/build.sh
new file mode 100755
index 0000000..d1fd2d4
--- /dev/null
+++ b/build.sh
@@ -0,0 +1,2 @@
+#!/bin/sh
+gcc -Wall -Wextra -Wno-pointer-sign $(xml2-config --cflags --libs) -o cite cite.c
diff --git a/cite.c b/cite.c
new file mode 100644
index 0000000..7db3749
--- /dev/null
+++ b/cite.c
@@ -0,0 +1,159 @@
+#include <libxml/parser.h>
+#include <string.h>
+
+typedef struct {
+	int id;
+	char title[256];
+	char og_title[256];
+	char directors[1024];
+	char actors[1024];
+	char prod_year[32];
+	char release_date[32];
+	char duration[32];
+	char main_genres[256];
+	char nationality[256];
+	char pitch[2048];
+	char poster[256];
+} Movie;
+
+Movie movies[256] = {0};
+
+static void print_style(void)
+{
+	puts("<link rel='stylesheet' type='text/css' href='https://kdx.re/theme.css'/>");
+}
+
+static void print_movie(const Movie *movie)
+{
+	if (movie->og_title[0] != '\0')
+		printf("<h2>%s</h2>\n", movie->og_title);
+	else
+		printf("<h2>%s</h2>\n", movie->title);
+	printf("<a href=\"%s\"><img src=\"%s\" alt=\"affiche\" /></a>\n",
+	       movie->poster, movie->poster);
+	printf("<ul>\n");
+	printf("<li>réalisateurs : %s</li>\n", movie->directors);
+	printf("<li>acteurs : %s</li>\n", movie->actors);
+	printf("<li>année de production : %s</li>\n", movie->prod_year);
+	printf("<li>date de sortie : %s</li>\n", movie->release_date);
+	printf("<li>durée : %s</li>\n", movie->duration);
+	printf("<li>genres : %s</li>\n", movie->main_genres);
+	printf("<li>nationalité : %s</li>\n", movie->nationality);
+	//printf("synopsis : %s\n", movie->pitch);
+	printf("</ul>\n");
+	printf("\n");
+}
+
+static void xfree(const void *ptr)
+{
+	if (ptr != NULL)
+		xmlFree((void *)ptr);
+}
+
+static int get_id(const xmlNode *node)
+{
+	const xmlChar *id = xmlGetProp(node, "id");
+	if (id == NULL)
+		return 0;
+	const int v = atoi(id);
+	xfree(id);
+	return v;
+}
+
+static const xmlNode *get_movie_node(const xmlNode *node)
+{
+	while (node != NULL && strcmp((const char *)node->name, "film") != 0)
+		node = node->children;
+	return node;
+}
+
+static Movie *get_movie(int id)
+{
+	Movie *movie = movies;
+	while (movie->id != id && movie->id != 0)
+		movie += 1;
+	return movie;
+}
+
+static int set_movie_base_fields(Movie *movie, const xmlNode *node)
+{
+	int err = 0;
+	const xmlChar *title = xmlGetProp(node, "titre");
+	const xmlChar *og_title = xmlGetProp(node, "titreoriginal");
+	const xmlChar *directors = xmlGetProp(node, "realisateurs");
+	const xmlChar *actors = xmlGetProp(node, "acteurs");
+	const xmlChar *prod_year = xmlGetProp(node, "anneeproduction");
+	const xmlChar *release_date = xmlGetProp(node, "datesortie");
+	const xmlChar *duration = xmlGetProp(node, "duree");
+	const xmlChar *main_genres = xmlGetProp(node, "genreprincipal");
+	const xmlChar *nationality = xmlGetProp(node, "nationalite");
+	const xmlChar *pitch = xmlGetProp(node, "synopsis");
+	const xmlChar *poster = xmlGetProp(node, "affichette");
+	if (title == NULL || og_title == NULL || directors == NULL ||
+	    actors == NULL || prod_year == NULL || release_date == NULL ||
+	    duration == NULL || main_genres == NULL || nationality == NULL ||
+	    pitch == NULL || poster == NULL) {
+		err = 1;
+		goto set_fields_panic;
+	}
+	strncpy(movie->title, title, sizeof(movie->title));
+	strncpy(movie->og_title, og_title, sizeof(movie->title));
+	strncpy(movie->directors, directors, sizeof(movie->directors));
+	strncpy(movie->actors, actors, sizeof(movie->actors));
+	strncpy(movie->prod_year, prod_year, sizeof(movie->prod_year));
+	strncpy(movie->release_date, release_date, sizeof(movie->release_date));
+	strncpy(movie->duration, duration, sizeof(movie->duration));
+	strncpy(movie->main_genres, main_genres, sizeof(movie->main_genres));
+	strncpy(movie->nationality, nationality, sizeof(movie->nationality));
+	strncpy(movie->pitch, pitch, sizeof(movie->pitch));
+	strncpy(movie->poster, poster, sizeof(movie->poster));
+set_fields_panic:
+	xfree(title);
+	xfree(og_title);
+	xfree(directors);
+	xfree(actors);
+	xfree(prod_year);
+	xfree(release_date);
+	xfree(duration);
+	xfree(main_genres);
+	xfree(nationality);
+	xfree(pitch);
+	xfree(poster);
+	return err;
+}
+
+int main(int argc, char **argv)
+{
+	if (argc != 2)
+		return 1;
+	xmlDoc *const document = xmlReadFile(argv[1], NULL, 0);
+	if (document == NULL) {
+		xmlCleanupParser();
+		return 1;
+	}
+	xmlNode *const root = xmlDocGetRootElement(document);
+	if (root == NULL) {
+		xmlCleanupParser();
+		xmlFreeDoc(document);
+		return 1;
+	}
+	for (const xmlNode *week = root->children; week != NULL; week = week->next) {
+		for (const xmlNode *mov = get_movie_node(week); mov != NULL; mov = mov->next) {
+			const int id = get_id(mov);
+			if (id == 0)
+				continue;
+			Movie *movie = get_movie(id);
+			if (movie->id == 0) {
+				movie->id = id;
+				set_movie_base_fields(movie, mov);
+			}
+		}
+	}
+	print_style();
+	printf("<h1>cinéma de la cité</h1>\n");
+	for (const Movie *movie = movies; movie->id != 0; movie++)
+		print_movie(movie);
+	xmlFreeDoc(document);
+	xmlCleanupParser();
+	return 0;
+}
diff --git a/scrap.sh b/scrap.sh
new file mode 100755
index 0000000..7a2943d
--- /dev/null
+++ b/scrap.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+./build.sh || exit 1
+curl -o seances.xml 'http://www.citebd.org/IMG/xml/allocineseances-4.xml' || exit 1
+./cite seances.xml >cite.html || exit 1
+firefox cite.html