Extend fz_archive for directories to count/list entries.

Also, move APIs to consistently use "fz_archive *dir"
rather than "fz_archive *zip" as this is felt to be
better.

include/mupdf/fitz/document.h

@@ -323,12 +323,13 @@ typedef void (fz_page_delete_link_fn)(fz_context *ctx, fz_page *page, fz_link *l
 	accel: fz_stream to read accelerator data from. May be
 	NULL. May be ignored.
 
-	zip: Archive to read associated content from (like images for
-	an html stream). Maybe NULL. May be ignored.
+	dir: 'Directory context' in which the document is loaded;
+	associated content from (like images for an html stream
+	will be loaded from this). Maybe NULL. May be ignored.
 
 	Pointer to opened document. Throws exception in case of error.
 */
-typedef fz_document *(fz_document_open_fn)(fz_context *ctx, fz_stream *stream, fz_stream *accel, fz_archive *zip);
+typedef fz_document *(fz_document_open_fn)(fz_context *ctx, fz_stream *stream, fz_stream *accel, fz_archive *dir);
 
 /**
 	Recognize a document type from
@@ -349,7 +350,7 @@ typedef int (fz_document_recognize_fn)(fz_context *ctx, const char *magic);
 	stream: stream contents to recognise (may be NULL if document is
 	a directory).
 
-	dir: directory from which stream is loaded.
+	dir: directory context from which stream is loaded.
 
 	Returns a number between 0 (not recognized) and 100
 	(fully recognized) based on how certain the recognizer

include/mupdf/fitz/story-writer.h

@@ -189,8 +189,8 @@ void fz_write_story(
  *      story content has been written to the device).
  *  pagefn_ref:
  *      Passed to pagefn().
- *  archive:
- *      NULL, or an archive to load images etc from.
+ *  dir:
+ *      NULL, or a directory context to load images etc from.
  */
 void fz_write_stabilized_story(
 		fz_context *ctx,
@@ -203,7 +203,7 @@ void fz_write_stabilized_story(
 		void *rectfn_ref,
 		fz_write_story_pagefn pagefn,
 		void *pagefn_ref,
-		fz_archive *archive
+		fz_archive *dir
 		);
 
 #endif

include/mupdf/fitz/story.h

@@ -56,11 +56,11 @@ typedef struct fz_story fz_story;
 
 	Passing a NULL buffer will be treated as an empty document.
 	Passing a NULL user_css will be treated as an empty CSS string.
-	A non-NULL archive will allow images etc to be loaded. The
+	A non-NULL dir will allow images etc to be loaded. The
 	story keeps its own reference, so the caller can drop its
 	reference after this call.
 */
-fz_story *fz_new_story(fz_context *ctx, fz_buffer *buf, const char *user_css, float em, fz_archive *archive);
+fz_story *fz_new_story(fz_context *ctx, fz_buffer *buf, const char *user_css, float em, fz_archive *dir);
 
 /*
 	Retrieve the warnings given from parsing this story.

include/mupdf/fitz/util.h

@@ -108,22 +108,22 @@ int fz_search_display_list(fz_context *ctx, fz_display_list *list, const char *n
 /**
 	Parse an SVG document into a display-list.
 */
-fz_display_list *fz_new_display_list_from_svg(fz_context *ctx, fz_buffer *buf, const char *base_uri, fz_archive *zip, float *w, float *h);
+fz_display_list *fz_new_display_list_from_svg(fz_context *ctx, fz_buffer *buf, const char *base_uri, fz_archive *dir, float *w, float *h);
 
 /**
 	Create a scalable image from an SVG document.
 */
-fz_image *fz_new_image_from_svg(fz_context *ctx, fz_buffer *buf, const char *base_uri, fz_archive *zip);
+fz_image *fz_new_image_from_svg(fz_context *ctx, fz_buffer *buf, const char *base_uri, fz_archive *dir);
 
 /**
 	Parse an SVG document into a display-list.
 */
-fz_display_list *fz_new_display_list_from_svg_xml(fz_context *ctx, fz_xml_doc *xmldoc, fz_xml *xml, const char *base_uri, fz_archive *zip, float *w, float *h);
+fz_display_list *fz_new_display_list_from_svg_xml(fz_context *ctx, fz_xml_doc *xmldoc, fz_xml *xml, const char *base_uri, fz_archive *dir, float *w, float *h);
 
 /**
 	Create a scalable image from an SVG document.
 */
-fz_image *fz_new_image_from_svg_xml(fz_context *ctx, fz_xml_doc *xmldoc, fz_xml *xml, const char *base_uri, fz_archive *zip);
+fz_image *fz_new_image_from_svg_xml(fz_context *ctx, fz_xml_doc *xmldoc, fz_xml *xml, const char *base_uri, fz_archive *dir);
 
 /**
 	Write image as a data URI (for HTML and SVG output).

include/mupdf/fitz/xml.h

@@ -57,7 +57,7 @@ fz_xml *fz_parse_xml_stream(fz_context *ctx, fz_stream *stream, int preserve_whi
 
 	preserve_white: whether to keep or delete all-whitespace nodes.
 */
-fz_xml *fz_parse_xml_archive_entry(fz_context *ctx, fz_archive *arch, const char *filename, int preserve_white);
+fz_xml *fz_parse_xml_archive_entry(fz_context *ctx, fz_archive *dir, const char *filename, int preserve_white);
 
 /**
 	Try and parse the contents of an archive entry into a tree of xml nodes.
@@ -67,7 +67,7 @@ fz_xml *fz_parse_xml_archive_entry(fz_context *ctx, fz_archive *arch, const char
 	Will return NULL if the archive entry can't be found. Otherwise behaves
 	the same as fz_parse_xml_archive_entry. May throw exceptions.
 */
-fz_xml *fz_try_parse_xml_archive_entry(fz_context *ctx, fz_archive *arch, const char *filename, int preserve_white);
+fz_xml *fz_try_parse_xml_archive_entry(fz_context *ctx, fz_archive *dir, const char *filename, int preserve_white);
 
 /**
 	Parse the contents of a buffer into a tree of XML nodes,

include/mupdf/html.h

@@ -44,9 +44,9 @@ typedef struct
 	int patch_mobi;
 } fz_htdoc_format_t;
 
-fz_document *fz_htdoc_open_document_with_buffer(fz_context *ctx, fz_archive *zip, fz_buffer *buf, const fz_htdoc_format_t *format);
+fz_document *fz_htdoc_open_document_with_buffer(fz_context *ctx, fz_archive *dir, fz_buffer *buf, const fz_htdoc_format_t *format);
 
-fz_document *fz_htdoc_open_document_with_stream_and_dir(fz_context *ctx, fz_stream *stm, fz_archive *zip, const fz_htdoc_format_t *format);
+fz_document *fz_htdoc_open_document_with_stream_and_dir(fz_context *ctx, fz_stream *stm, fz_archive *dir, const fz_htdoc_format_t *format);
 
 
 

source/cbz/mucbz.c

@@ -258,7 +258,7 @@ cbz_lookup_metadata(fz_context *ctx, fz_document *doc_, const char *key, char *b
 }
 
 static fz_document *
-cbz_open_document(fz_context *ctx, fz_stream *file, fz_stream *accel, fz_archive *zip)
+cbz_open_document(fz_context *ctx, fz_stream *file, fz_stream *accel, fz_archive *dir)
 {
 	cbz_document *doc = fz_new_derived_document(ctx, cbz_document);
 
@@ -269,7 +269,10 @@ cbz_open_document(fz_context *ctx, fz_stream *file, fz_stream *accel, fz_archive
 
 	fz_try(ctx)
 	{
-		doc->arch = fz_open_archive_with_stream(ctx, file);
+		if (file)
+			doc->arch = fz_open_archive_with_stream(ctx, file);
+		else
+			doc->arch = fz_keep_archive(ctx, dir);
 		cbz_create_page_list(ctx, doc);
 	}
 	fz_catch(ctx)
@@ -318,26 +321,27 @@ cbz_recognize_doc_content(fz_context *ctx, fz_stream *stream, fz_archive *dir)
 	fz_var(arch);
 	fz_var(ret);
 
-	/* FIXME: Maybe consider: if stream == NULL then use dir as archive.
-	 * This would enable us to open directories of unpacked cbz's. Is
-	 * this a good thing, or is it too permissive? */
-
-	if (stream == NULL)
-		return 0;
-
 	fz_try(ctx)
 	{
-		arch = fz_try_open_archive_with_stream(ctx, stream);
-		if (arch == NULL)
-			break;
+		if (stream == NULL)
+			arch = fz_keep_archive(ctx, dir);
+		else
+		{
+			arch = fz_try_open_archive_with_stream(ctx, stream);
+			if (arch == NULL)
+				break;
+		}
 
 		/* If it's an archive, and we can find at least one plausible page
 		 * then we can open it as a cbz. */
 		count = fz_count_archive_entries(ctx, arch);
 		for (i = 0; i < count && ret == 0; i++)
 		{
 			const char *name = fz_list_archive_entry(ctx, arch, i);
-			const char *ext = name ? strrchr(name, '.') : NULL;
+			const char *ext;
+			if (name == NULL)
+				continue;
+			ext = strrchr(name, '.');
 			if (ext)
 			{
 				for (k = 0; cbz_ext_list[k]; k++)

source/cbz/muimg.c

@@ -172,7 +172,7 @@ img_lookup_metadata(fz_context *ctx, fz_document *doc_, const char *key, char *b
 }
 
 static fz_document *
-img_open_document(fz_context *ctx, fz_stream *file, fz_stream *accel, fz_archive *zip)
+img_open_document(fz_context *ctx, fz_stream *file, fz_stream *accel, fz_archive *dir)
 {
 	img_document *doc = fz_new_derived_document(ctx, img_document);
 

source/fitz/directory.c

@@ -27,21 +27,32 @@
 #include <sys/stat.h>
 
 #ifdef _MSC_VER
+#include <windows.h>
 #define stat _stat
+#else
+#include <dirent.h>
 #endif
 
 typedef struct
 {
 	fz_archive super;
 
 	char *path;
+
+	int max_entries;
+	int num_entries;
+	char **entries;
 } fz_directory;
 
 static void drop_directory(fz_context *ctx, fz_archive *arch)
 {
 	fz_directory *dir = (fz_directory *) arch;
+	int i;
 
 	fz_free(ctx, dir->path);
+	for (i = 0; i < dir->num_entries; i++)
+		fz_free(ctx, dir->entries[i]);
+	fz_free(ctx, dir->entries);
 }
 
 static fz_stream *open_dir_entry(fz_context *ctx, fz_archive *arch, const char *name)
@@ -85,25 +96,150 @@ fz_is_directory(fz_context *ctx, const char *path)
 	return S_ISDIR(info.st_mode);
 }
 
+static int
+count_dir_entries(fz_context *ctx, fz_archive *arch)
+{
+	fz_directory *dir = (fz_directory *) arch;
+
+	return dir->num_entries;
+}
+
+const char *
+list_dir_entry(fz_context *ctx, fz_archive *arch, int n)
+{
+	fz_directory *dir = (fz_directory *) arch;
+
+	if (n < 0 || n >= dir->num_entries)
+		return NULL;
+
+	return dir->entries[n];
+}
+
 fz_archive *
 fz_open_directory(fz_context *ctx, const char *path)
 {
 	fz_directory *dir;
+#ifdef _MSC_VER
+	WCHAR *wpath = NULL;
+	size_t z = 3;
+	HANDLE h = NULL;
+	WIN32_FIND_DATAW dw;
+
+	fz_var(wpath);
+	fz_var(h);
+#else
+	DIR *dp;
+	struct dirent *ep;
+
+	fz_var(dp);
+#endif
 
 	if (!fz_is_directory(ctx, path))
 		fz_throw(ctx, FZ_ERROR_FORMAT, "'%s' is not a directory", path);
 
 	dir = fz_new_derived_archive(ctx, NULL, fz_directory);
 	dir->super.format = "dir";
+	dir->super.count_entries = count_dir_entries;
+	dir->super.list_entry = list_dir_entry;
 	dir->super.has_entry = has_dir_entry;
 	dir->super.read_entry = read_dir_entry;
 	dir->super.open_entry = open_dir_entry;
 	dir->super.drop_archive = drop_directory;
 
 	fz_try(ctx)
 	{
+#ifdef _MSC_VER
+		char const *p = path;
+		WCHAR *w;
+		while (*p)
+		{
+			int rune;
+			p += fz_chartorune(&rune, p);
+			if (rune >= 0x10000 || rune < 0)
+				fz_throw(ctx, FZ_ERROR_GENERIC, "Unrepresentable UTF-8 char in directory name");
+			z++;
+		}
+		w = wpath = fz_malloc(ctx, z * sizeof(WCHAR));
+		p = path;
+		while (*p)
+		{
+			int rune;
+			p += fz_chartorune(&rune, p);
+			*w++ = rune;
+		}
+		w[0] = '\\';
+		w[1] = '*';
+		w[2] = 0;
+
+		/* Now enumerate the paths. */
+		h = FindFirstFileW(wpath, &dw);
+		if (h == INVALID_HANDLE_VALUE)
+			break;
+
+		do
+		{
+			char *u;
+
+			if (dir->max_entries == dir->num_entries)
+			{
+				int newmax = dir->max_entries * 2;
+				if (newmax == 0)
+					newmax = 32;
+
+				dir->entries = fz_realloc(ctx, dir->entries, sizeof(*dir->entries) * newmax);
+				dir->max_entries = newmax;
+			}
+
+			/* Count the len as utf-8. */
+			w = dw.cFileName;
+			z = 1;
+			while (*w)
+				z += fz_runelen(*w++);
+
+			u = dir->entries[dir->num_entries] = fz_malloc(ctx, z);
+			dir->num_entries++;
+
+			/* Copy the name across. */
+			w = dw.cFileName;
+			while (*w)
+				u += fz_runetochar(u, *w++);
+			*u = 0;
+		}
+		while (FindNextFileW(h, &dw));
+#else
+		dp = opendir(path);
+		if (dp == NULL)
+			break;
+
+		while ((ep = readdir(dp)) != NULL)
+		{
+			if (dir->max_entries == dir->num_entries)
+			{
+				int newmax = dir->max_entries * 2;
+				if (newmax == 0)
+					newmax = 32;
+
+				dir->entries = fz_realloc(ctx, dir->entries, sizeof(*dir->entries) * newmax);
+				dir->max_entries = newmax;
+			}
+
+			dir->entries[dir->num_entries] = fz_strdup(ctx, ep->d_name);
+			dir->num_entries++;
+		}
+#endif
 		dir->path = fz_strdup(ctx, path);
 	}
+	fz_always(ctx)
+	{
+#ifdef _MSC_VER
+		fz_free(ctx, wpath);
+		if (h)
+			(void)FindClose(h);
+#else
+		if (dp)
+			(void)closedir(dp);
+#endif
+	}
 	fz_catch(ctx)
 	{
 		fz_drop_archive(ctx, &dir->super);

source/fitz/document.c

@@ -249,9 +249,9 @@ fz_open_document_with_stream(fz_context *ctx, const char *magic, fz_stream *stre
 }
 
 fz_document *
-fz_open_document_with_stream_and_dir(fz_context *ctx, const char *magic, fz_stream *stream, fz_archive *zip)
+fz_open_document_with_stream_and_dir(fz_context *ctx, const char *magic, fz_stream *stream, fz_archive *dir)
 {
-	return fz_open_accelerated_document_with_stream_and_dir(ctx, magic, stream, NULL, zip);
+	return fz_open_accelerated_document_with_stream_and_dir(ctx, magic, stream, NULL, dir);
 }
 
 fz_document *
@@ -288,12 +288,12 @@ fz_open_accelerated_document(fz_context *ctx, const char *filename, const char *
 	if (fz_is_directory(ctx, filename))
 	{
 		/* Cannot accelerate directories, currently. */
-		fz_archive *zip = fz_open_directory(ctx, filename);
+		fz_archive *dir = fz_open_directory(ctx, filename);
 
 		fz_try(ctx)
-			doc = fz_open_accelerated_document_with_stream_and_dir(ctx, filename, NULL, NULL, zip);
+			doc = fz_open_accelerated_document_with_stream_and_dir(ctx, filename, NULL, NULL, dir);
 		fz_always(ctx)
-			fz_drop_archive(ctx, zip);
+			fz_drop_archive(ctx, dir);
 		fz_catch(ctx)
 			fz_rethrow(ctx);