fixed document sequence parsing

12 files changed, 580 insertions, 10 deletions

diff --git a/apps/samshow.c b/apps/samshow.c
index c8351dc1..e9ce86f1 100644
--- a/apps/samshow.c
+++ b/apps/samshow.c
@@ -1,6 +1,33 @@
 #include "fitz.h"
 #include "samus.h"
 
+void showfixdocseq(sa_package *pack, char *part)
+{
+	fz_error *error;
+	sa_fixdocseq *seq;
+
+	error = sa_loadfixdocseq(&seq, pack, part);
+	if (error)
+		fz_abort(error);
+
+	sa_debugfixdocseq(seq);
+
+	sa_dropfixdocseq(seq);
+}
+
+void showreach(sa_package *pack, sa_relation *rels)
+{
+	while (rels)
+	{
+		if (!strcmp(rels->type, SA_REL_FIXEDREPRESENTATION))
+		{
+			if (!rels->external)
+				showfixdocseq(pack, rels->target);
+		}
+		rels = rels->next;
+	}
+}
+
 int runpack(int argc, char **argv)
 {
 	fz_error *error;
@@ -14,18 +41,23 @@ int runpack(int argc, char **argv)
 		fz_abort(error);
 
 	sa_debugpackage(pack);
+	printf("\n");
 
-	printf("--- root ---\n");
 	error = sa_loadrelations(&rels, pack, "/");
 	if (error)
 		fz_abort(error);
 	sa_debugrelations(rels);
-	sa_droprelations(rels);
 	printf("\n");
 
+	if (argc == 2)
+	{
+		showreach(pack, rels);
+		return 0;
+	}
+
 	for (i = 2; i < argc; i++)
 	{
-		printf("--- %s ---\n", argv[i]);
+		printf("part %s\n", argv[i]);
 
 		s = sa_typepart(pack, argv[i]);
 		if (!s)
@@ -121,7 +153,7 @@ int runtiff(int argc, char **argv)
 	error = sa_readtiff(file);
 	if (error)
 		fz_abort(error);
-	
+
 	fz_dropstream(file);
 
 	return 0;
diff --git a/base/Jamfile b/base/Jamfile
index 0436cffd..91bde812 100644
--- a/base/Jamfile
+++ b/base/Jamfile
@@ -8,6 +8,7 @@ Library libbase :
 	base_memory.c
 	base_rect.c
 	base_rune.c
+	base_cleanname.c
 	;
 
 if $(NEED_GETOPT)  = yes { Library libbase : util_getopt.c ; }
diff --git a/base/base_cleanname.c b/base/base_cleanname.c
new file mode 100644
index 00000000..54764630
--- /dev/null
+++ b/base/base_cleanname.c
@@ -0,0 +1,52 @@
+/*
+ * In place, rewrite name to compress multiple /, eliminate ., and process ..
+ */
+
+#define SEP(x)	((x)=='/' || (x) == 0)
+
+char *
+cleanname(char *name)
+{
+	char *p, *q, *dotdot;
+	int rooted;
+
+	rooted = name[0] == '/';
+
+	/*
+	 * invariants:
+	 *	p points at beginning of path element we're considering.
+	 *	q points just past the last path element we wrote (no slash).
+	 *	dotdot points just past the point where .. cannot backtrack
+	 *		any further (no slash).
+	 */
+	p = q = dotdot = name+rooted;
+	while(*p) {
+		if(p[0] == '/')	/* null element */
+			p++;
+		else if(p[0] == '.' && SEP(p[1]))
+			p += 1;	/* don't count the separator in case it is nul */
+		else if(p[0] == '.' && p[1] == '.' && SEP(p[2])) {
+			p += 2;
+			if(q > dotdot) {	/* can backtrack */
+				while(--q > dotdot && *q != '/')
+					;
+			} else if(!rooted) {	/* /.. is / but ./../ is .. */
+				if(q != name)
+					*q++ = '/';
+				*q++ = '.';
+				*q++ = '.';
+				dotdot = q;
+			}
+		} else {	/* real path element */
+			if(q != name+rooted)
+				*q++ = '/';
+			while((*q = *p) != '/' && *q != 0)
+				p++, q++;
+		}
+	}
+	if(q == name)	/* empty string is really ``.'' */
+		*q++ = '.';
+	*q = '\0';
+	return name;
+}
+
diff --git a/include/fitz/runtime.h b/include/fitz/runtime.h
index e52e3c1d..feeb7ffc 100644
--- a/include/fitz/runtime.h
+++ b/include/fitz/runtime.h
@@ -24,11 +24,13 @@
 
 #define STRIDE(n, bcp) (((bpc) * (n) + 7) / 8)
 
+/* plan9 stuff for utf-8 and path munging */
 int chartorune(int *rune, char *str);
 int runetochar(char *str, int *rune);
 int runelen(long c);
 int runenlen(int *r, int nrune);
 int fullrune(char *str, int n);
+char *cleanname(char *name);
 
 typedef struct fz_error_s fz_error;
 
diff --git a/include/samus.h b/include/samus.h
index e9c841f7..a7b39482 100644
--- a/include/samus.h
+++ b/include/samus.h
@@ -20,3 +20,6 @@
 #include "samus/xml.h"
 #include "samus/pack.h"
 
+#include "samus/names.h"
+#include "samus/fixdoc.h"
+
diff --git a/include/samus/fixdoc.h b/include/samus/fixdoc.h
new file mode 100644
index 00000000..1b0caad1
--- /dev/null
+++ b/include/samus/fixdoc.h
@@ -0,0 +1,13 @@
+/*
+ * FixedDocumentSequence -- the list of pages
+ */
+
+typedef struct sa_fixdocseq_s sa_fixdocseq;
+
+fz_error *sa_loadfixdocseq(sa_fixdocseq **seqp, sa_package *pack, char *part);
+void sa_debugfixdocseq(sa_fixdocseq *seq);
+void sa_dropfixdocseq(sa_fixdocseq *seq);
+
+int sa_getpagecount(sa_fixdocseq *seq);
+char *sa_getpagepart(sa_fixdocseq *seq, int idx);
+
diff --git a/include/samus/misc.h b/include/samus/misc.h
index db4bfc78..50170a04 100644
--- a/include/samus/misc.h
+++ b/include/samus/misc.h
@@ -2,5 +2,6 @@
  * Misc utility functions that Samus needs.
  */
 
+char *sa_resolvepath(char *dst, char *base, char *part, int dstlen);
 int sa_strcmp(char *s0, char *s1);
 
diff --git a/include/samus/names.h b/include/samus/names.h
new file mode 100644
index 00000000..03a0e505
--- /dev/null
+++ b/include/samus/names.h
@@ -0,0 +1,96 @@
+/*
+ * Various long string constants
+ */
+
+/*
+ * XML Namespaces
+ */
+
+#define SA_NS_CONTENTTYPES \
+	"http://schemas.microsoft.com/metro/2005/02/content-types"
+#define SA_NS_COREPROPERTIES \
+	"http://schemas.microsoft.com/metro/2005/02/md/core-properties"
+#define SA_NS_SIGNATURES \
+	"http://schemas.microsoft.com/metro/2005/02/digsig"
+#define SA_NS_RELATIONSHIPS \
+	"http://schemas.microsoft.com/metro/2005/02/relationships"
+#define SA_NS_VERSIONING \
+	"http://schemas.microsoft.com/winfx/markup-compatibility/2005"
+
+#define SA_NS_DISCARDCONTROL \
+	"http://schemas.microsoft.com/metro/2005/02/rp/discard-control"
+#define SA_NS_FIXEDDOCUMENT \
+	"http://schemas.microsoft.com/metro/2005/02/rp"
+#define SA_NS_FIXEDDOCUMENTSEQUENCE \
+	"http://schemas.microsoft.com/metro/2005/02/rp"
+#define SA_NS_FIXEDPAGE \
+	"http://schemas.microsoft.com/metro/2005/02/rp"
+#define SA_NS_PRINTSCHEMAFRAMEWORK \
+	"http://schemas.microsoft.com/windows/2003/08/printing/printschemaframework"
+#define SA_NS_PRINTSCHEMAKEYWORDS \
+	"http://schemas.microsoft.com/windows/2003/08/printing/printschemakeywords"
+#define SA_NS_RESOURCEDICTIONARY \
+	"http://schemas.microsoft.com/metro/2005/02/rp/resourcedictionary-key"
+
+/*
+ * Content types
+ */
+
+#define SA_TYPE_COREPROPERTIES \
+	"application/vnd.ms-metro.core-properties+xml"
+#define SA_TYPE_CERTIFICATE \
+	"application/vnd.ms-metro.ds-certificate"
+#define SA_TYPE_ORIGIN \
+	"application/vnd.ms-metro.ds-origin"
+#define SA_TYPE_SIGNATURE \
+	"application/vnd.ms-metro.ds-xmlsignature+xml"
+#define SA_TYPE_RELATIONSHIPS \
+	"application/vnd.ms-metro.relationships+xml"
+
+#define SA_TYPE_FIXEDDOCUMENT \
+	"application/vnd.ms-metro.rp-fixeddocument+xml"
+#define SA_TYPE_FIXEDDOCUMENTSEQUENCE \
+	"application/vnd.ms-metro.rp-fixeddocumentsequence+xml"
+#define SA_TYPE_FIXEDPAGE \
+	"application/vnd.ms-metro.rp-fixedpage+xml"
+#define SA_TYPE_FONT \
+	"application/vnd.ms-opentype"
+#define SA_TYPE_OBFUSCATEDFONT \
+	"application/vnd.ms-metro.obfuscated-opentype"
+#define SA_TYPE_PRINTTICKET \
+	"application/vnd.ms-printing.printticket+xml"
+#define SA_TYPE_JPEGIMAGE \
+	"image/jpeg"
+#define SA_TYPE_PNGIMAGE \
+	"image/png"
+#define SA_TYPE_TIFFIMAGE \
+	"image/tiff"
+
+/*
+ * Relationship types
+ */
+
+#define SA_REL_COREPROPERTIES \
+	"http://schemas.microsoft.com/metro/2005/02/md/core-properties"
+#define SA_REL_SIGNATURE \
+	"http://schemas.microsoft.com/metro/2005/02/ds/signature"
+#define SA_REL_CERTIFICATE \
+	"http://schemas.microsoft.com/metro/2005/02/ds/certificate"
+#define SA_REL_ORIGIN \
+	"http://schemas.microsoft.com/metro/2005/02/ds/origin"
+#define SA_REL_THUMBNAIL \
+	"http://schemas.microsoft.com/metro/2005/02/md/thumbnail"
+
+#define SA_REL_ANNOTATIONS \
+	"http://schemas.microsoft.com/metro/2005/02/rp/annotations"
+#define SA_REL_DISCARDCONTROL \
+	"http://schemas.microsoft.com/metro/2005/02/rp/discard-control"
+#define SA_REL_PRINTTICKET \
+	"http://schemas.microsoft.com/metro/2005/02/rp/printticket"
+#define SA_REL_FIXEDREPRESENTATION \
+	"http://schemas.microsoft.com/metro/2005/02/rp/fixedrepresentation"
+#define SA_REL_REQUIREDRESOURCE \
+	"http://schemas.microsoft.com/metro/2005/02/rp/required-resource"
+#define SA_REL_RESTRICTEDFONT \
+	"http://schemas.microsoft.com/metro/2005/02/rp/restricted-font"
+
diff --git a/samus/Jamfile b/samus/Jamfile
index dde4a934..45d11dd5 100644
--- a/samus/Jamfile
+++ b/samus/Jamfile
@@ -1,10 +1,14 @@
 SubDir TOP samus ;
 
 Library libsamus :
+
 	sa_misc.c
 	sa_zip.c
 	sa_xml.c
 	sa_tiff.c
 	sa_pack.c
+
+	sa_fixdoc.c
+
 	;
 
diff --git a/samus/sa_fixdoc.c b/samus/sa_fixdoc.c
new file mode 100644
index 00000000..5e69064f
--- /dev/null
+++ b/samus/sa_fixdoc.c
@@ -0,0 +1,327 @@
+/*
+ * FixedDocumentSequence and FixedDocument parts.
+ */
+
+#include "fitz.h"
+#include "samus.h"
+
+typedef struct sa_fixdoc_s sa_fixdoc;
+typedef struct sa_fixpage_s sa_fixpage;
+
+struct sa_fixdocseq_s
+{
+	int count;	/* pages */
+	sa_fixdoc *docs;
+};
+
+struct sa_fixdoc_s
+{
+	int count;
+	sa_fixpage *pages;
+	sa_fixdoc *next;
+};
+
+struct sa_fixpage_s
+{
+	char *part;
+	int width;
+	int height;
+	/* char **linktargets; */
+	sa_fixpage *next;
+};
+
+/*
+ * Debugging
+ */
+
+static void
+sa_debugfixdoc(sa_fixdoc *doc, int i)
+{
+	sa_fixpage *page = doc->pages;
+	printf("  FixedDocument\n  {\n");
+	while (page)
+	{
+		printf("    FixedPage %d w=%d h=%d %s\n", i,
+				page->width, page->height, page->part);
+		page = page->next;
+		i ++;
+	}
+	printf("  }\n");
+}
+
+void
+sa_debugfixdocseq(sa_fixdocseq *seq)
+{
+	sa_fixdoc *doc = seq->docs;
+	int i = 0;
+	printf("FixedDocumentSequence count=%d\n{\n", seq->count);
+	while (doc)
+	{
+		sa_debugfixdoc(doc, i);
+		i += doc->count;
+		doc = doc->next;
+	}
+	printf("}\n");
+}
+
+/*
+ * Free data structures
+ */
+
+static void
+sa_dropfixpage(sa_fixpage *page)
+{
+	sa_fixpage *next;
+	while (page)
+	{
+		next = page->next;
+		fz_free(page->part);
+		fz_free(page);
+		page = next;
+	}
+}
+
+static void
+sa_dropfixdoc(sa_fixdoc *doc)
+{
+	sa_fixdoc *next;
+	while (doc)
+	{
+		next = doc->next;
+		if (doc->pages)
+			sa_dropfixpage(doc->pages);
+		fz_free(doc);
+		doc = next;
+	}
+}
+
+void
+sa_dropfixdocseq(sa_fixdocseq *seq)
+{
+	sa_dropfixdoc(seq->docs);
+	fz_free(seq);
+}
+
+/*
+ * Load FixedDocument
+ */
+static fz_error *
+sa_loadfixdoc(sa_fixdoc **docp, sa_package *pack, char *part)
+{
+	fz_error *error;
+	fz_stream *stm;
+	sa_xmlparser *parser;
+	sa_xmlitem *item;
+	sa_fixdoc *doc;
+	sa_fixpage *page;
+	sa_fixpage *last;
+	char buf[1024];
+
+	page = nil;
+	last = nil;
+
+	error = sa_openpart(&stm, pack, part);
+	if (error)
+		return error;
+
+	doc = fz_malloc(sizeof(sa_fixdoc));
+	if (!doc)
+	{
+		error = fz_outofmem;
+		goto cleanupstm;
+	}
+
+	doc->count = 0;
+	doc->pages = nil;
+	doc->next = nil;
+
+	error = sa_openxml(&parser, stm, 0);
+	if (error)
+		goto cleanupdoc;
+
+	item = sa_xmlnext(parser);
+	while (item)
+	{
+		if (!strcmp(sa_xmlname(item), "FixedDocument"))
+		{
+			sa_xmldown(parser);
+			item = sa_xmlnext(parser);
+			while (item)
+			{
+				if (!strcmp(sa_xmlname(item), "PageContent"))
+				{
+					char *src = sa_xmlatt(item, "Source");
+					char *w = sa_xmlatt(item, "Width");
+					char *h = sa_xmlatt(item, "Height");
+
+					if (!w) w = "0";
+					if (!h) h = "0";
+
+					if (src)
+					{
+						sa_resolvepath(buf, part, src, sizeof buf);
+						page = fz_malloc(sizeof(sa_fixpage));
+						if (!page)
+						{
+							error = fz_outofmem;
+							goto cleanupxml;
+						}
+
+						page->part = fz_strdup(buf);
+						if (!page->part)
+						{
+							fz_free(page);
+							error = fz_outofmem;
+							goto cleanupxml;
+						}
+
+						page->width = atoi(w);
+						page->height = atoi(h);
+
+						if (last)
+							last->next = page;
+						else
+							doc->pages = page;
+						doc->count ++;
+						last = page;
+					}
+				}
+				item = sa_xmlnext(parser);
+			}
+			sa_xmlup(parser);
+		}
+		item = sa_xmlnext(parser);
+	}
+
+	sa_closexml(parser);
+	fz_dropstream(stm);
+	*docp = doc;
+	return nil;
+
+cleanupxml:
+	sa_closexml(parser);
+cleanupdoc:
+	sa_dropfixdoc(doc);
+cleanupstm:
+	fz_dropstream(stm);
+	return error;
+}
+
+/*
+ * Load FixedDocumentSequence
+ */
+fz_error *
+sa_loadfixdocseq(sa_fixdocseq **seqp, sa_package *pack, char *part)
+{
+	fz_error *error;
+	fz_stream *stm;
+	sa_xmlparser *parser;
+	sa_xmlitem *item;
+	sa_fixdocseq *seq;
+	sa_fixdoc *doc;
+	sa_fixdoc *last;
+	char buf[1024];
+
+	seq = nil;
+	last = nil;
+
+	error = sa_openpart(&stm, pack, part);
+	if (error)
+		return error;
+
+	seq = fz_malloc(sizeof(sa_fixdocseq));
+	if (!seq)
+	{
+		error = fz_outofmem;
+		goto cleanupstm;
+	}
+
+	seq->count = 0;
+	seq->docs = nil;
+
+	error = sa_openxml(&parser, stm, 0);
+	if (error)
+		goto cleanupseq;
+
+	item = sa_xmlnext(parser);
+	while (item)
+	{
+		if (!strcmp(sa_xmlname(item), "FixedDocumentSequence"))
+		{
+			sa_xmldown(parser);
+			item = sa_xmlnext(parser);
+			while (item)
+			{
+				if (!strcmp(sa_xmlname(item), "DocumentReference"))
+				{
+					char *src = sa_xmlatt(item, "Source");
+					if (src)
+					{
+						sa_resolvepath(buf, part, src, sizeof buf);
+						error = sa_loadfixdoc(&doc, pack, buf);
+						if (error)
+							goto cleanupxml;
+						if (last)
+							last->next = doc;
+						else
+							seq->docs = doc;
+						seq->count += doc->count;
+						last = doc;
+					}
+				}
+				item = sa_xmlnext(parser);
+			}
+			sa_xmlup(parser);
+		}
+		item = sa_xmlnext(parser);
+	}
+
+	sa_closexml(parser);
+	fz_dropstream(stm);
+	*seqp = seq;
+	return nil;
+
+cleanupxml:
+	sa_closexml(parser);
+cleanupseq:
+	sa_dropfixdocseq(seq);
+cleanupstm:
+	fz_dropstream(stm);
+	return error;
+}
+
+/*
+ * Accessors
+ */
+
+int
+sa_getpagecount(sa_fixdocseq *seq)
+{
+	return seq->count;
+}
+
+char *
+sa_getpagepart(sa_fixdocseq *seq, int idx)
+{
+	sa_fixdoc *doc = seq->docs;
+	int cur = 0;
+
+	if (idx < 0 || idx >= seq->count)
+		return nil;
+
+	while (doc)
+	{
+		sa_fixpage *page = doc->pages;
+		while (page)
+		{
+			if (idx == cur)
+				return page->part;
+			page = page->next;
+			cur ++;
+		}
+		doc = doc->next;
+	}
+
+	return nil;
+}
+
diff --git a/samus/sa_misc.c b/samus/sa_misc.c
index 109e3693..0f71b4a2 100644
--- a/samus/sa_misc.c
+++ b/samus/sa_misc.c
@@ -2,6 +2,32 @@
 #include "samus.h"
 
 /*
+ * Create a new canonical absolute path given both
+ * a base url (current directory) and a relative or absolute name.
+ */
+char *
+sa_resolvepath(char *dst, char *baseurl, char *parturl, int dstlen)
+{
+	char *end;
+
+	if (parturl[0] == '/')
+	{
+		strlcpy(dst, parturl, dstlen);
+	}
+	else
+	{
+		strlcpy(dst, baseurl, dstlen);
+		end = strrchr(dst, '/');
+		if (end)
+			end[1] = 0;
+		strlcat(dst, parturl, dstlen);
+		cleanname(dst);
+	}
+
+	return dst;
+}
+
+/*
  * Test part names for equivalence.
  *
  * What we *should* do here (according to the spec) is...
diff --git a/samus/sa_pack.c b/samus/sa_pack.c
index 74640ebd..4c65ba38 100644
--- a/samus/sa_pack.c
+++ b/samus/sa_pack.c
@@ -215,16 +215,20 @@ sa_debugpackage(sa_package *pack)
 	sa_default *def;
 	sa_override *ovr;
 
-	printf("package {\n");
+	printf("package\n{\n");
 
 	if (pack->zip)
 		sa_debugzip(pack->zip);
 
+	printf("  defaults\n  {\n");
 	for (def = pack->defaults; def; def = def->next)
-		printf("default %s %s ;\n", def->extension, def->mimetype);
+		printf("    %-8s %s\n", def->extension, def->mimetype);
+	printf("  }\n");
 
+	printf("  overrides\n  {\n");
 	for (ovr = pack->overrides; ovr; ovr = ovr->next)
-		printf("override %s %s ;\n", ovr->partname, ovr->mimetype);
+		printf("    %s\n        %s\n", ovr->partname, ovr->mimetype);
+	printf("  }\n");
 
 	printf("}\n");
 }
@@ -256,6 +260,7 @@ sa_loadrelations(sa_relation **relsp, sa_package *pack, char *partname)
 	int len;
 	char *sep;
 	char *relsname;
+	char buf[1024];
 
 	if (partname[0] != '/')
 		return fz_throw("ioerror: invalid part name: %s", partname);
@@ -318,9 +323,15 @@ sa_loadrelations(sa_relation **relsp, sa_package *pack, char *partname)
 						if (!newrel) { error = fz_outofmem; goto cleanupxml; }
 						newrel->external = !strcmp(mode, "External");
 						newrel->id = fz_strdup(id);
-						newrel->target = fz_strdup(target);
 						newrel->type = fz_strdup(type);
 						newrel->next = rels;
+						if (newrel->external)
+							newrel->target = fz_strdup(target);
+						else
+						{
+							sa_resolvepath(buf, partname, target, sizeof buf);
+							newrel->target = fz_strdup(buf);
+						}
 						rels = newrel;
 					}
 				}
@@ -349,12 +360,14 @@ cleanupname:
 void
 sa_debugrelations(sa_relation *rel)
 {
+	printf("relations\n{\n");
 	while (rel)
 	{
-		printf("relation %s\n", rel->type);
-		printf("    %s%s\n", rel->external ? "external " : "", rel->target);
+		printf("  %s\n", rel->type);
+		printf("      %s%s\n", rel->external ? "external " : "", rel->target);
 		rel = rel->next;
 	}
+	printf("}\n");
 }
 
 void