From 0cd7f2e8397e2caf47f3d46ef635358fa0ba194c Mon Sep 17 00:00:00 2001
From: Sebastian Rasmussen <sebras@gmail.com>
Date: Tue, 28 Mar 2017 20:14:53 +0800
Subject: Move extension/mimetype detection to common function.

A document handler normally only exposes a list of extensions and
mimetypes. Only formats that use some kind of extra detection mechnism
need to supply a recognize() callback, such as xps that can handle
.xps-files unpacked into a directory.
---
 source/cbz/mucbz.c     | 41 ++++++++++-----------
 source/cbz/muimg.c     | 82 ++++++++++++++++++++++--------------------
 source/cbz/mutiff.c    | 31 ++++++++--------
 source/fitz/document.c | 97 +++++++++++++++++++++++++++++++-------------------
 source/gprf/gprf-doc.c | 28 ++++++++-------
 source/html/epub-doc.c | 22 ++++++++----
 source/html/html-doc.c | 38 +++++++++++---------
 source/pdf/pdf-xref.c  | 28 +++++++--------
 source/svg/svg-doc.c   | 28 ++++++++-------
 source/xps/xps-doc.c   | 31 +++++++++-------
 10 files changed, 238 insertions(+), 188 deletions(-)

(limited to 'source')

diff --git a/source/cbz/mucbz.c b/source/cbz/mucbz.c
index a557a9c7..3a19c174 100644
--- a/source/cbz/mucbz.c
+++ b/source/cbz/mucbz.c
@@ -217,28 +217,29 @@ cbz_open_document_with_stream(fz_context *ctx, fz_stream *file)
 	return &doc->super;
 }
 
-static int
-cbz_recognize(fz_context *ctx, const char *magic)
-{
-	char *ext = strrchr(magic, '.');
-	if ((ext && !fz_strcasecmp(ext, ".cbz")) || !strcmp(magic, "cbz") ||
-			!strcmp(magic, "application/x-cbz"))
-		return 100;
-	if ((ext && !fz_strcasecmp(ext, ".zip")) || !strcmp(magic, "zip") ||
-			!strcmp(magic, "application/zip"))
-		return 100;
-	if ((ext && !fz_strcasecmp(ext, ".tar")) || !strcmp(magic, "tar") ||
-			!strcmp(magic, "application/x-tar"))
-		return 100;
-	if ((ext && !fz_strcasecmp(ext, ".cbt")) || !strcmp(magic, "cbt") ||
-			!strcmp(magic, "application/x-cbt"))
-		return 100;
-	return 0;
-}
+static const char *cbz_extensions[] =
+{
+	"cbt",
+	"cbz",
+	"tar",
+	"zip",
+	NULL
+};
+
+static const char *cbz_mimetypes[] =
+{
+	"application/x-cbt",
+	"application/x-cbz",
+	"application/x-tar",
+	"application/zip",
+	NULL
+};
 
 fz_document_handler cbz_document_handler =
 {
-	cbz_recognize,
 	NULL,
-	cbz_open_document_with_stream
+	NULL,
+	cbz_open_document_with_stream,
+	cbz_extensions,
+	cbz_mimetypes
 };
diff --git a/source/cbz/muimg.c b/source/cbz/muimg.c
index 4c590f52..088dee22 100644
--- a/source/cbz/muimg.c
+++ b/source/cbz/muimg.c
@@ -130,49 +130,53 @@ img_open_document_with_stream(fz_context *ctx, fz_stream *stm)
 	return &doc->super;
 }
 
-static int
-img_recognize(fz_context *doc, const char *magic)
+static const char *img_extensions[] =
 {
-	char *ext = strrchr(magic, '.');
+	"bmp",
+	"gif",
+	"hdp",
+	"j2k",
+	"jfif",
+	"jfif-tbnl",
+	"jp2",
+	"jpe",
+	"jpeg",
+	"jpg",
+	"jpx",
+	"jxr",
+	"pam",
+	"pbm",
+	"pgm",
+	"png",
+	"pnm",
+	"ppm",
+	"wdp",
+	NULL
+};
 
-	if (ext)
-	{
-		if (!fz_strcasecmp(ext, ".png") || !fz_strcasecmp(ext, ".jpg") ||
-			!fz_strcasecmp(ext, ".jpeg") || !fz_strcasecmp(ext, ".jfif") ||
-			!fz_strcasecmp(ext, ".jfif-tbnl") || !fz_strcasecmp(ext, ".jpe") ||
-			!fz_strcasecmp(ext, ".gif") || !fz_strcasecmp(ext, ".bmp") ||
-			!fz_strcasecmp(ext, ".jpx") || !fz_strcasecmp(ext, ".jp2") ||
-			!fz_strcasecmp(ext, ".j2k") || !fz_strcasecmp(ext, ".wdp") ||
-			!fz_strcasecmp(ext, ".hdp") || !fz_strcasecmp(ext, ".jxr") ||
-			!fz_strcasecmp(ext, ".pbm") || !fz_strcasecmp(ext, ".pgm") ||
-			!fz_strcasecmp(ext, ".ppm") || !fz_strcasecmp(ext, ".pam") ||
-			!fz_strcasecmp(ext, ".pnm"))
-			return 100;
-	}
-	if (!strcmp(magic, "png") || !strcmp(magic, "image/png") ||
-		!strcmp(magic, "jpg") || !strcmp(magic, "image/jpeg") ||
-		!strcmp(magic, "jpeg") || !strcmp(magic, "image/pjpeg") ||
-		!strcmp(magic, "jpe") || !strcmp(magic, "jfif") ||
-		!strcmp(magic, "gif") || !strcmp(magic, "image/gif") ||
-		!strcmp(magic, "bmp") || !strcmp(magic, "image/bmp") ||
-		!strcmp(magic, "jpx") || !strcmp(magic, "image/jpx") ||
-		!strcmp(magic, "jp2") || !strcmp(magic, "image/jp2") ||
-		!strcmp(magic, "j2k") || !strcmp(magic, "wdp") ||
-		!strcmp(magic, "hdp") || !strcmp(magic, "image/vnd.ms-photo") ||
-		!strcmp(magic, "jxr") || !strcmp(magic, "image/jxr") ||
-		!strcmp(magic, "pbm") || !strcmp(magic, "image/x-portable-bitmap") ||
-		!strcmp(magic, "pgm") || !strcmp(magic, "image/x-portable-greymap") ||
-		!strcmp(magic, "ppm") || !strcmp(magic, "image/x-portable-pixmap") ||
-		!strcmp(magic, "pam") || !strcmp(magic, "image/x-portable-arbitrarymap") ||
-		!strcmp(magic, "pnm"))
-		return 100;
-
-	return 0;
-}
+static const char *img_mimetypes[] =
+{
+	"image/bmp",
+	"image/gif",
+	"image/jp2",
+	"image/jpeg",
+	"image/jpx",
+	"image/jxr",
+	"image/pjpeg",
+	"image/png",
+	"image/vnd.ms-photo",
+	"image/x-portable-arbitrarymap",
+	"image/x-portable-bitmap",
+	"image/x-portable-greymap",
+	"image/x-portable-pixmap",
+	NULL
+};
 
 fz_document_handler img_document_handler =
 {
-	img_recognize,
 	NULL,
-	img_open_document_with_stream
+	NULL,
+	img_open_document_with_stream,
+	img_extensions,
+	img_mimetypes
 };
diff --git a/source/cbz/mutiff.c b/source/cbz/mutiff.c
index 637f4e3a..06c7a837 100644
--- a/source/cbz/mutiff.c
+++ b/source/cbz/mutiff.c
@@ -143,26 +143,25 @@ tiff_open_document_with_stream(fz_context *ctx, fz_stream *file)
 	return &doc->super;
 }
 
-static int
-tiff_recognize(fz_context *doc, const char *magic)
+static const char *tiff_extensions[] =
 {
-	char *ext = strrchr(magic, '.');
-
-	if (ext)
-	{
-		if (!fz_strcasecmp(ext, ".tiff") || !fz_strcasecmp(ext, ".tif"))
-			return 100;
-	}
-	if (!strcmp(magic, "tif") || !strcmp(magic, "image/tiff") ||
-		!strcmp(magic, "tiff") || !strcmp(magic, "image/x-tiff"))
-		return 100;
+	"tif",
+	"tiff",
+	NULL
+};
 
-	return 0;
-}
+static const char *tiff_mimetypes[] =
+{
+	"image/tiff",
+	"image/x-tiff",
+	NULL
+};
 
 fz_document_handler tiff_document_handler =
 {
-	tiff_recognize,
 	NULL,
-	tiff_open_document_with_stream
+	NULL,
+	tiff_open_document_with_stream,
+	tiff_extensions,
+	tiff_mimetypes
 };
diff --git a/source/fitz/document.c b/source/fitz/document.c
index 8bf8d348..8f3163c4 100644
--- a/source/fitz/document.c
+++ b/source/fitz/document.c
@@ -63,25 +63,51 @@ void fz_register_document_handler(fz_context *ctx, const fz_document_handler *ha
 	dc->handler[dc->count++] = handler;
 }
 
-fz_document *
-fz_open_document_with_stream(fz_context *ctx, const char *magic, fz_stream *stream)
+static const fz_document_handler *
+fz_recognize_document(fz_context *ctx, const char *magic)
 {
-	int i, score;
-	int best_i, best_score;
 	fz_document_handler_context *dc;
-
-	if (magic == NULL || stream == NULL)
-		fz_throw(ctx, FZ_ERROR_GENERIC, "no document to open");
+	int i, best_score, best_i;
+	const char *ext, *needle;
 
 	dc = ctx->handler;
 	if (dc->count == 0)
 		fz_throw(ctx, FZ_ERROR_GENERIC, "No document handlers registered");
 
-	best_i = -1;
+	ext = strrchr(magic, '.');
+	if (ext)
+		needle = ext + 1;
+	else
+		needle = magic;
+
 	best_score = 0;
+	best_i = -1;
+
 	for (i = 0; i < dc->count; i++)
 	{
-		score = dc->handler[i]->recognize(ctx, magic);
+		int score = 0;
+		const char **entry;
+
+		if (dc->handler[i]->recognize)
+			score = dc->handler[i]->recognize(ctx, magic);
+
+		if (!ext)
+		{
+			for (entry = &dc->handler[i]->mimetypes[0]; *entry; entry++)
+				if (!fz_strcasecmp(needle, *entry) && score < 100)
+				{
+					score = 100;
+					break;
+				}
+		}
+
+		for (entry = &dc->handler[i]->extensions[0]; *entry; entry++)
+			if (!fz_strcasecmp(needle, *entry) && score < 100)
+			{
+				score = 100;
+				break;
+			}
+
 		if (best_score < score)
 		{
 			best_score = score;
@@ -89,50 +115,49 @@ fz_open_document_with_stream(fz_context *ctx, const char *magic, fz_stream *stre
 		}
 	}
 
-	if (best_i >= 0)
-		return dc->handler[best_i]->open_with_stream(ctx, stream);
+	if (best_i < 0)
+		return NULL;
 
-	fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find document handler for file type: %s", magic);
+	return dc->handler[best_i];
+}
+
+
+fz_document *
+fz_open_document_with_stream(fz_context *ctx, const char *magic, fz_stream *stream)
+{
+	const fz_document_handler *handler;
+
+	if (magic == NULL || stream == NULL)
+		fz_throw(ctx, FZ_ERROR_GENERIC, "no document to open");
+
+	handler = fz_recognize_document(ctx, magic);
+	if (!handler)
+		fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find document handler for file type: %s", magic);
+
+	return handler->open_with_stream(ctx, stream);
 }
 
 fz_document *
 fz_open_document(fz_context *ctx, const char *filename)
 {
-	int i, score;
-	int best_i, best_score;
-	fz_document_handler_context *dc;
+	const fz_document_handler *handler;
 	fz_stream *file;
 	fz_document *doc;
 
 	if (filename == NULL)
 		fz_throw(ctx, FZ_ERROR_GENERIC, "no document to open");
 
-	dc = ctx->handler;
-	if (dc->count == 0)
-		fz_throw(ctx, FZ_ERROR_GENERIC, "No document handlers registered");
-
-	best_i = -1;
-	best_score = 0;
-	for (i = 0; i < dc->count; i++)
-	{
-		score = dc->handler[i]->recognize(ctx, filename);
-		if (best_score < score)
-		{
-			best_score = score;
-			best_i = i;
-		}
-	}
-
-	if (best_i < 0)
-		fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find document handler for file: '%s'", filename);
+	handler = fz_recognize_document(ctx, filename);
+	if (!handler)
+		fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find document handler for file: %s", filename);
 
-	if (dc->handler[best_i]->open)
-		return dc->handler[best_i]->open(ctx, filename);
+	if (handler->open)
+		return handler->open(ctx, filename);
 
 	file = fz_open_file(ctx, filename);
 
 	fz_try(ctx)
-		doc = dc->handler[best_i]->open_with_stream(ctx, file);
+		doc = handler->open_with_stream(ctx, file);
 	fz_always(ctx)
 		fz_drop_stream(ctx, file);
 	fz_catch(ctx)
diff --git a/source/gprf/gprf-doc.c b/source/gprf/gprf-doc.c
index faeb8e2f..53816a15 100644
--- a/source/gprf/gprf-doc.c
+++ b/source/gprf/gprf-doc.c
@@ -990,22 +990,24 @@ gprf_open_document(fz_context *ctx, const char *filename)
 	return doc;
 }
 
-static int
-gprf_recognize(fz_context *doc, const char *magic)
-{
-	char *ext = strrchr(magic, '.');
-	if (ext)
-		if (!fz_strcasecmp(ext, ".gproof"))
-			return 100;
-	if (!strcmp(magic, "application/ghostproof"))
-		return 100;
-	return 0;
-}
+static const char *gprf_extensions[] =
+{
+	"gproof",
+	NULL
+};
+
+static const char *gprf_mimetypes[] =
+{
+	"application/x-ghostproof",
+	NULL
+};
 
 fz_document_handler gprf_document_handler =
 {
-	gprf_recognize,
+	NULL,
 	gprf_open_document,
-	gprf_open_document_with_stream
+	gprf_open_document_with_stream,
+	gprf_extensions,
+	gprf_mimetypes
 };
 #endif
diff --git a/source/html/epub-doc.c b/source/html/epub-doc.c
index 44903d99..c43ef8b1 100644
--- a/source/html/epub-doc.c
+++ b/source/html/epub-doc.c
@@ -519,20 +519,28 @@ epub_open_document(fz_context *ctx, const char *filename)
 static int
 epub_recognize(fz_context *doc, const char *magic)
 {
-	char *ext = strrchr(magic, '.');
-	if (ext)
-		if (!fz_strcasecmp(ext, ".epub"))
-			return 100;
 	if (strstr(magic, "META-INF/container.xml") || strstr(magic, "META-INF\\container.xml"))
 		return 200;
-	if (!strcmp(magic, "application/epub+zip"))
-		return 100;
 	return 0;
 }
 
+static const char *epub_extensions[] =
+{
+	"epub",
+	NULL
+};
+
+static const char *epub_mimetypes[] =
+{
+	"application/epub+zip",
+	NULL
+};
+
 fz_document_handler epub_document_handler =
 {
 	epub_recognize,
 	epub_open_document,
-	epub_open_document_with_stream
+	epub_open_document_with_stream,
+	epub_extensions,
+	epub_mimetypes
 };
diff --git a/source/html/html-doc.c b/source/html/html-doc.c
index ba765350..1cf3dca5 100644
--- a/source/html/html-doc.c
+++ b/source/html/html-doc.c
@@ -197,26 +197,30 @@ htdoc_open_document(fz_context *ctx, const char *filename)
 	return (fz_document*)doc;
 }
 
-static int
-htdoc_recognize(fz_context *doc, const char *magic)
-{
-	char *ext = strrchr(magic, '.');
+static const char *htdoc_extensions[] =
+{
+	"fb2",
+	"htm",
+	"html",
+	"xhtml",
+	"xml",
+	NULL
+};
 
-	if (ext)
-	{
-		if (!fz_strcasecmp(ext, ".xml") || !fz_strcasecmp(ext, ".xhtml") ||
-				!fz_strcasecmp(ext, ".html") || !fz_strcasecmp(ext, ".htm") ||
-				!fz_strcasecmp(ext, ".fb2"))
-			return 100;
-	}
-	if (!strcmp(magic, "application/html+xml") || !strcmp(magic, "application/xml") || !strcmp(magic, "text/xml"))
-		return 100;
-	return 0;
-}
+static const char *htdoc_mimetypes[] =
+{
+	"application/html+xml",
+	"application/x-fictionbook",
+	"application/xml",
+	"text/xml",
+	NULL
+};
 
 fz_document_handler html_document_handler =
 {
-	htdoc_recognize,
+	NULL,
 	htdoc_open_document,
-	htdoc_open_document_with_stream
+	htdoc_open_document_with_stream,
+	htdoc_extensions,
+	htdoc_mimetypes
 };
diff --git a/source/pdf/pdf-xref.c b/source/pdf/pdf-xref.c
index 14da300d..e24da9a1 100644
--- a/source/pdf/pdf-xref.c
+++ b/source/pdf/pdf-xref.c
@@ -2738,27 +2738,25 @@ pdf_document *pdf_create_document(fz_context *ctx)
 	return doc;
 }
 
-int
-pdf_recognize(fz_context *doc, const char *magic)
+static const char *pdf_extensions[] =
 {
-	char *ext = strrchr(magic, '.');
-
-	if (ext)
-	{
-		if (!fz_strcasecmp(ext, ".pdf"))
-			return 100;
-	}
-	if (!strcmp(magic, "pdf") || !strcmp(magic, "application/pdf"))
-		return 100;
+	"pdf",
+	NULL
+};
 
-	return 1;
-}
+static const char *pdf_mimetypes[] =
+{
+	"application/pdf",
+	NULL
+};
 
 fz_document_handler pdf_document_handler =
 {
-	pdf_recognize,
+	NULL,
 	(fz_document_open_fn *) pdf_open_document,
-	(fz_document_open_with_stream_fn *) pdf_open_document_with_stream
+	(fz_document_open_with_stream_fn *) pdf_open_document_with_stream,
+	pdf_extensions,
+	pdf_mimetypes
 };
 
 void pdf_mark_xref(fz_context *ctx, pdf_document *doc)
diff --git a/source/svg/svg-doc.c b/source/svg/svg-doc.c
index c7f8add8..2a9f061e 100644
--- a/source/svg/svg-doc.c
+++ b/source/svg/svg-doc.c
@@ -121,16 +121,6 @@ svg_open_document_with_stream(fz_context *ctx, fz_stream *file)
 	return doc;
 }
 
-static int
-svg_recognize(fz_context *ctx, const char *magic)
-{
-	char *ext = strrchr(magic, '.');
-	if (ext && !fz_strcasecmp(ext, ".svg"))
-		return 100;
-	if (!strcmp(magic, "svg") || !strcmp(magic, "image/svg+xml"))
-		return 100;
-	return 0;
-}
 
 fz_display_list *
 fz_new_display_list_from_svg(fz_context *ctx, fz_buffer *buf, float *w, float *h)
@@ -170,9 +160,23 @@ fz_new_image_from_svg(fz_context *ctx, fz_buffer *buf)
 	return image;
 }
 
+static const char *svg_extensions[] =
+{
+	"svg",
+	NULL
+};
+
+static const char *svg_mimetypes[] =
+{
+	"image/svg+xml",
+	NULL
+};
+
 fz_document_handler svg_document_handler =
 {
-	svg_recognize,
 	NULL,
-	svg_open_document_with_stream
+	NULL,
+	svg_open_document_with_stream,
+	svg_extensions,
+	svg_mimetypes
 };
diff --git a/source/xps/xps-doc.c b/source/xps/xps-doc.c
index 5c91a591..23ecf17c 100644
--- a/source/xps/xps-doc.c
+++ b/source/xps/xps-doc.c
@@ -465,26 +465,31 @@ xps_load_page(fz_context *ctx, xps_document *doc, int number)
 static int
 xps_recognize(fz_context *ctx, const char *magic)
 {
-	char *ext = strrchr(magic, '.');
-
-	if (ext)
-	{
-		if (!fz_strcasecmp(ext, ".xps") || !fz_strcasecmp(ext, ".oxps"))
-			return 100;
-	}
 	if (strstr(magic, "/_rels/.rels") || strstr(magic, "\\_rels\\.rels"))
 		return 100;
-	if (!strcmp(magic, "xps") || !strcmp(magic, "oxps") ||
-		!strcmp(magic, "application/vnd.ms-xpsdocument") ||
-		!strcmp(magic, "application/xps") ||
-		!strcmp(magic, "application/oxps"))
-		return 100;
 	return 0;
 }
 
+static const char *xps_extensions[] =
+{
+	"oxps",
+	"xps",
+	NULL
+};
+
+static const char *xps_mimetypes[] =
+{
+	"application/oxps",
+	"application/vnd.ms-xpsdocument",
+	"application/xps",
+	NULL
+};
+
 fz_document_handler xps_document_handler =
 {
 	xps_recognize,
 	(fz_document_open_fn *) xps_open_document,
-	(fz_document_open_with_stream_fn *) xps_open_document_with_stream
+	(fz_document_open_with_stream_fn *) xps_open_document_with_stream,
+	xps_extensions,
+	xps_mimetypes
 };
-- 
cgit v1.2.3