Convert all strings to UTF-8 while dumping XML. (related RhBug: 920795)

diff --git a/src/xml_dump.c b/src/xml_dump.c
index 16148d1..a230ea0 100644 (file)
--- a/src/xml_dump.c
+++ b/src/xml_dump.c
@@ -43,6 +43,83 @@ cr_xml_dump_cleanup()
 
 
 void
+cr_latin1_to_utf8(const unsigned char *in, unsigned char *out)
+{
+    // http://stackoverflow.com/questions/4059775/convert-iso-8859-1-strings-to-utf-8-in-c-c/4059934#4059934
+    // This function converts latin1 to utf8 in effective and thread-safe way.
+    while (*in) {
+        if (*in<128) {
+            *out++=*in++;
+        } else if (*in<192) {
+            // Found latin1 (iso-8859-1) control code.
+            // The string is probably misencoded cp-1252 and not a real latin1.
+            // Just skip this character.
+            in++;
+            continue;
+        } else {
+            *out++=0xc2+(*in>0xbf);
+            *out++=(*in++&0x3f)+0x80;
+        }
+    }
+    *out = '\0';
+}
+
+xmlNodePtr
+cr_xmlNewTextChild(xmlNodePtr parent,
+                   xmlNsPtr ns,
+                   const xmlChar *name,
+                   const xmlChar *orig_content)
+{
+    int free_content = 0;
+    xmlChar *content;
+    xmlNodePtr child;
+
+    if (!orig_content) {
+        content = "";
+    } else if (xmlCheckUTF8(orig_content)) {
+        content = (xmlChar *) orig_content;
+    } else {
+        size_t len = strlen(orig_content);
+        content = malloc(sizeof(xmlChar)*len*2 + 1);
+        cr_latin1_to_utf8(orig_content, content);
+        free_content = 1;
+    }
+
+    child = xmlNewTextChild(parent, ns, name, content);
+
+    if (free_content)
+        free(content);
+
+    return child;
+}
+
+xmlAttrPtr
+cr_xmlNewProp(xmlNodePtr node, const xmlChar *name, const xmlChar *orig_content)
+{
+    int free_content = 0;
+    xmlChar *content;
+    xmlAttrPtr attr;
+
+    if (!orig_content) {
+        content = "";
+    } else if (xmlCheckUTF8(orig_content)) {
+        content = (xmlChar *) orig_content;
+    } else {
+        size_t len = strlen(orig_content);
+        content = malloc(sizeof(xmlChar)*len*2 + 1);
+        cr_latin1_to_utf8(orig_content, content);
+        free_content = 1;
+    }
+
+    attr = xmlNewProp(node, name, content);
+
+    if (free_content)
+        free(content);
+
+    return attr;
+}
+
+void
 cr_xml_dump_files(xmlNodePtr node, cr_Package *package, int primary)
 {
     if (!node || !package->files) {
@@ -85,15 +162,15 @@ cr_xml_dump_files(xmlNodePtr node, cr_Package *package, int primary)
         // ************************************
 
         xmlNodePtr file_node;
-        file_node = xmlNewTextChild(node,
-                                    NULL,
-                                    BAD_CAST "file",
-                                    BAD_CAST fullname);
+        file_node = cr_xmlNewTextChild(node,
+                                       NULL,
+                                       BAD_CAST "file",
+                                       BAD_CAST fullname);
         g_free(fullname);
 
         // Write type (skip type if type value is empty of "file")
         if (entry->type && entry->type[0] != '\0' && strcmp(entry->type, "file")) {
-            xmlNewProp(file_node, BAD_CAST "type", BAD_CAST entry->type);
+            cr_xmlNewProp(file_node, BAD_CAST "type", BAD_CAST entry->type);
         }
     }
 }

diff --git a/src/xml_dump.h b/src/xml_dump.h
index 4f65a19..128d181 100644 (file)
--- a/src/xml_dump.h
+++ b/src/xml_dump.h
@@ -105,6 +105,13 @@ char *cr_xml_dump_other(cr_Package *package);
  */
 struct cr_XmlStruct cr_xml_dump(cr_Package *package);
 
+/** Prepare string to xml dump.
+ * If string is not utf8 it is converted (source encoding is supposed to be
+ * iso-8859-1).
+ * TODO
+ */
+void cr_latin1_to_utf8(const unsigned char *in, unsigned char *out);
+
 /** @} */
 
 #ifdef __cplusplus

diff --git a/src/xml_dump_filelists.c b/src/xml_dump_filelists.c
index ef89aa7..9150e87 100644 (file)
--- a/src/xml_dump_filelists.c
+++ b/src/xml_dump_filelists.c
@@ -40,16 +40,13 @@ cr_xml_dump_filelists_items(xmlNodePtr root, cr_Package *package)
     ************************************/
 
     // Add pkgid attribute
-    xmlNewProp(root, BAD_CAST "pkgid",
-               BAD_CAST ((package->pkgId) ? package->pkgId : ""));
+    cr_xmlNewProp(root, BAD_CAST "pkgid", BAD_CAST package->pkgId);
 
     // Add name attribute
-    xmlNewProp(root, BAD_CAST "name",
-               BAD_CAST ((package->name) ? package->name : ""));
+    cr_xmlNewProp(root, BAD_CAST "name", BAD_CAST package->name);
 
     // Add arch attribute
-    xmlNewProp(root, BAD_CAST "arch",
-               BAD_CAST ((package->arch) ? package->arch : ""));
+    cr_xmlNewProp(root, BAD_CAST "arch", BAD_CAST package->arch);
 
 
     /***********************************
@@ -62,16 +59,13 @@ cr_xml_dump_filelists_items(xmlNodePtr root, cr_Package *package)
     version = xmlNewChild(root, NULL, BAD_CAST "version", NULL);
 
     // Write version attribute epoch
-    xmlNewProp(version, BAD_CAST "epoch",
-               BAD_CAST ((package->epoch) ? package->epoch : ""));
+    cr_xmlNewProp(version, BAD_CAST "epoch", BAD_CAST package->epoch);
 
     // Write version attribute ver
-    xmlNewProp(version, BAD_CAST "ver",
-               BAD_CAST ((package->version) ? package->version : ""));
+    cr_xmlNewProp(version, BAD_CAST "ver", BAD_CAST package->version);
 
     // Write version attribute rel
-    xmlNewProp(version, BAD_CAST "rel",
-               BAD_CAST ((package->release) ? package->release : ""));
+    cr_xmlNewProp(version, BAD_CAST "rel", BAD_CAST package->release);
 
 
     // Files dump

diff --git a/src/xml_dump_internal.h b/src/xml_dump_internal.h
index c0dd40c..4a0c7d1 100644 (file)
--- a/src/xml_dump_internal.h
+++ b/src/xml_dump_internal.h
@@ -25,6 +25,7 @@ extern "C" {
 #endif
 
 #include "package.h"
+#include <libxml/tree.h>
 
 /**
  * Dump files from the package and append them to the node as childrens.
@@ -35,6 +36,21 @@ extern "C" {
  */
 void cr_xml_dump_files(xmlNodePtr node, cr_Package *package, int primary);
 
+/**
+ * Createrepo wrapper over libxml xmlNewTextChild.
+ * It allows content to be NULL and non UTF-8 (if content is no UTF8
+ * then iso-8859-1 is assumed).
+ */
+xmlNodePtr cr_xmlNewTextChild(xmlNodePtr parent,
+                              xmlNsPtr ns,
+                              const xmlChar *name,
+                              const xmlChar *content);
+
+/** TODO */
+xmlAttrPtr cr_xmlNewProp(xmlNodePtr node,
+                         const xmlChar *name,
+                         const xmlChar *value);
+
 #ifdef __cplusplus
 }
 #endif

diff --git a/src/xml_dump_other.c b/src/xml_dump_other.c
index a73bb81..74f41dc 100644 (file)
--- a/src/xml_dump_other.c
+++ b/src/xml_dump_other.c
@@ -23,9 +23,11 @@
 #include <string.h>
 #include <libxml/encoding.h>
 #include <libxml/xmlwriter.h>
+#include <libxml/xmlsave.h>
 #include "logging.h"
 #include "package.h"
 #include "xml_dump.h"
+#include "xml_dump_internal.h"
 
 #define FORMAT_XML      1
 #define FORMAT_LEVEL    0
@@ -55,15 +57,13 @@ cr_xml_dump_other_changelog(xmlNodePtr root, cr_Package *package)
         xmlNodePtr changelog;
 
         // Add changelog element
-        changelog = xmlNewTextChild(root,
-                                    NULL,
-                                    BAD_CAST "changelog",
-                                    BAD_CAST ((entry->changelog) ? entry->changelog : ""));
+        changelog = cr_xmlNewTextChild(root,
+                                       NULL,
+                                       BAD_CAST "changelog",
+                                       BAD_CAST entry->changelog);
 
         // Write param author
-        xmlNewProp(changelog,
-                   BAD_CAST "author",
-                   BAD_CAST ((entry->author) ? entry->author : ""));
+        cr_xmlNewProp(changelog, BAD_CAST "author", BAD_CAST entry->author);
 
         // Write param date
         char date_str[DATE_MAX_LEN];
@@ -81,16 +81,13 @@ cr_xml_dump_other_items(xmlNodePtr root, cr_Package *package)
     ************************************/
 
     // Add pkgid attribute
-    xmlNewProp(root, BAD_CAST "pkgid",
-               BAD_CAST ((package->pkgId) ? package->pkgId : ""));
+    cr_xmlNewProp(root, BAD_CAST "pkgid", BAD_CAST package->pkgId);
 
     // Add name attribute
-    xmlNewProp(root, BAD_CAST "name",
-               BAD_CAST ((package->name) ? package->name : ""));
+    cr_xmlNewProp(root, BAD_CAST "name", BAD_CAST package->name);
 
     // Add arch attribute
-    xmlNewProp(root, BAD_CAST "arch",
-               BAD_CAST ((package->arch) ? package->arch : ""));
+    cr_xmlNewProp(root, BAD_CAST "arch", BAD_CAST package->arch);
 
 
     /***********************************
@@ -103,16 +100,13 @@ cr_xml_dump_other_items(xmlNodePtr root, cr_Package *package)
     version = xmlNewChild(root, NULL, BAD_CAST "version", NULL);
 
     // Write version attribute epoch
-    xmlNewProp(version, BAD_CAST "epoch",
-               BAD_CAST ((package->epoch) ? package->epoch : ""));
+    xmlNewProp(version, BAD_CAST "epoch", BAD_CAST package->epoch);
 
     // Write version attribute ver
-    xmlNewProp(version, BAD_CAST "ver",
-               BAD_CAST ((package->version) ? package->version : ""));
+    xmlNewProp(version, BAD_CAST "ver", BAD_CAST package->version);
 
     // Write version attribute rel
-    xmlNewProp(version, BAD_CAST "rel",
-               BAD_CAST ((package->release) ? package->release : ""));
+    xmlNewProp(version, BAD_CAST "rel", BAD_CAST package->release);
 
 
     // Changelog dump
@@ -142,6 +136,8 @@ cr_xml_dump_other(cr_Package *package)
         return NULL;
     }
     // Seems to be little bit faster than xmlDocDumpFormatMemory
+//    xmlSaveCtxtPtr savebuf = xmlSaveToBuffer(buf, NULL, XML_SAVE_FORMAT|XML_SAVE_NO_DECL);
+//    xmlSaveTree(savebuf, root);
     xmlNodeDump(buf, NULL, root, FORMAT_LEVEL, FORMAT_XML);
     assert(buf->content);
     result = g_strndup((char *) buf->content, (buf->use+1));

diff --git a/src/xml_dump_primary.c b/src/xml_dump_primary.c
index b141526..50390e0 100644 (file)
--- a/src/xml_dump_primary.c
+++ b/src/xml_dump_primary.c
@@ -93,27 +93,27 @@ cr_xml_dump_primary_dump_pco(xmlNodePtr root, cr_Package *package, int pcotype)
 
 
         /***********************************
-        Element: entry
+         Element: entry
         ************************************/
 
         xmlNodePtr entry_node;
 
         entry_node = xmlNewChild(pcor_node, NULL, BAD_CAST "rpm:entry", NULL);
-        xmlNewProp(entry_node, BAD_CAST "name", BAD_CAST entry->name);
+        cr_xmlNewProp(entry_node, BAD_CAST "name", BAD_CAST entry->name);
 
         if (entry->flags && entry->flags[0] != '\0') {
-            xmlNewProp(entry_node, BAD_CAST "flags", BAD_CAST entry->flags);
+            cr_xmlNewProp(entry_node, BAD_CAST "flags", BAD_CAST entry->flags);
 
             if (entry->epoch && entry->epoch[0] != '\0') {
-                xmlNewProp(entry_node, BAD_CAST "epoch", BAD_CAST entry->epoch);
+                cr_xmlNewProp(entry_node, BAD_CAST "epoch", BAD_CAST entry->epoch);
             }
 
             if (entry->version && entry->version[0] != '\0') {
-                xmlNewProp(entry_node, BAD_CAST "ver", BAD_CAST entry->version);
+                cr_xmlNewProp(entry_node, BAD_CAST "ver", BAD_CAST entry->version);
             }
 
             if (entry->release && entry->release[0] != '\0') {
-                xmlNewProp(entry_node, BAD_CAST "rel", BAD_CAST entry->release);
+                cr_xmlNewProp(entry_node, BAD_CAST "rel", BAD_CAST entry->release);
             }
         }
 
@@ -141,16 +141,14 @@ cr_xml_dump_primary_base_items(xmlNodePtr root, cr_Package *package)
      Element: name
     ************************************/
 
-    xmlNewTextChild(root, NULL,  BAD_CAST "name",
-                    BAD_CAST ((package->name) ? package->name : ""));
+    cr_xmlNewTextChild(root, NULL,  BAD_CAST "name", BAD_CAST package->name);
 
 
     /***********************************
      Element: arch
     ************************************/
 
-    xmlNewTextChild(root, NULL,  BAD_CAST "arch",
-                    BAD_CAST ((package->arch) ? package->arch : ""));
+    cr_xmlNewTextChild(root, NULL,  BAD_CAST "arch", BAD_CAST package->arch);
 
 
     /***********************************
@@ -163,16 +161,13 @@ cr_xml_dump_primary_base_items(xmlNodePtr root, cr_Package *package)
     version = xmlNewChild(root, NULL, BAD_CAST "version", NULL);
 
     // Write version attribute epoch
-    xmlNewProp(version, BAD_CAST "epoch",
-               BAD_CAST ((package->epoch) ? package->epoch : ""));
+    cr_xmlNewProp(version, BAD_CAST "epoch", BAD_CAST package->epoch);
 
     // Write version attribute ver
-    xmlNewProp(version, BAD_CAST "ver",
-               BAD_CAST ((package->version) ? package->version : ""));
+    cr_xmlNewProp(version, BAD_CAST "ver", BAD_CAST package->version);
 
     // Write version attribute rel
-    xmlNewProp(version, BAD_CAST "rel",
-               BAD_CAST ((package->release) ? package->release : ""));
+    cr_xmlNewProp(version, BAD_CAST "rel", BAD_CAST package->release);
 
 
     /***********************************
@@ -181,12 +176,13 @@ cr_xml_dump_primary_base_items(xmlNodePtr root, cr_Package *package)
 
     xmlNodePtr checksum;
 
-    checksum = xmlNewTextChild(root, NULL,  BAD_CAST "checksum",
-                               BAD_CAST ((package->pkgId) ? package->pkgId : ""));
+    checksum = cr_xmlNewTextChild(root,
+                                  NULL,
+                                  BAD_CAST "checksum",
+                                  BAD_CAST package->pkgId);
 
     // Write checksum attribute checksum_type
-    xmlNewProp(checksum, BAD_CAST "type",
-               BAD_CAST ((package->checksum_type) ? package->checksum_type : ""));
+    cr_xmlNewProp(checksum, BAD_CAST "type", BAD_CAST package->checksum_type);
 
     // Write checksum attribute pkgid
     xmlNewProp(checksum, BAD_CAST "pkgid", BAD_CAST "YES");
@@ -196,32 +192,31 @@ cr_xml_dump_primary_base_items(xmlNodePtr root, cr_Package *package)
      Element: summary
     ************************************/
 
-    xmlNewTextChild(root, NULL,  BAD_CAST "summary",
-                    BAD_CAST ((package->summary) ? package->summary : ""));
+    cr_xmlNewTextChild(root, NULL,  BAD_CAST "summary",
+                       BAD_CAST package->summary);
 
 
     /***********************************
     Element: description
     ************************************/
 
-    xmlNewTextChild(root, NULL,  BAD_CAST "description",
-                    BAD_CAST ((package->description) ? package->description : ""));
+    cr_xmlNewTextChild(root, NULL,  BAD_CAST "description",
+                       BAD_CAST package->description);
 
 
     /***********************************
      Element: packager
     ************************************/
 
-    xmlNewTextChild(root, NULL,  BAD_CAST "packager",
-                    BAD_CAST ((package->rpm_packager) ? package->rpm_packager : ""));
+    cr_xmlNewTextChild(root, NULL,  BAD_CAST "packager",
+                       BAD_CAST package->rpm_packager);
 
 
     /***********************************
      Element: url
     ************************************/
 
-    xmlNewTextChild(root, NULL,  BAD_CAST "url",
-                    BAD_CAST ((package->url) ? package->url : ""));
+    cr_xmlNewTextChild(root, NULL,  BAD_CAST "url", BAD_CAST package->url);
 
 
     /***********************************
@@ -279,14 +274,13 @@ cr_xml_dump_primary_base_items(xmlNodePtr root, cr_Package *package)
 
     // Write location attribute base
     if (package->location_base && package->location_base[0] != '\0') {
-        xmlNewProp(location,
-                   BAD_CAST "xml:base",
-                   BAD_CAST package->location_base);
+        cr_xmlNewProp(location,
+                      BAD_CAST "xml:base",
+                      BAD_CAST package->location_base);
     }
 
     // Write location attribute href
-    xmlNewProp(location, BAD_CAST "href",
-               BAD_CAST ((package->location_href) ? package->location_href : ""));
+    cr_xmlNewProp(location, BAD_CAST "href", BAD_CAST package->location_href);
 
 
     /***********************************
@@ -302,40 +296,40 @@ cr_xml_dump_primary_base_items(xmlNodePtr root, cr_Package *package)
      Element: license
     ************************************/
 
-    xmlNewTextChild(format, NULL,  BAD_CAST "rpm:license",
-                    BAD_CAST ((package->rpm_license) ? package->rpm_license : ""));
+    cr_xmlNewTextChild(format, NULL,  BAD_CAST "rpm:license",
+                       BAD_CAST package->rpm_license);
 
 
     /***********************************
      Element: vendor
     ************************************/
 
-    xmlNewTextChild(format, NULL,  BAD_CAST "rpm:vendor",
-                    BAD_CAST ((package->rpm_vendor) ? package->rpm_vendor : ""));
+    cr_xmlNewTextChild(format, NULL,  BAD_CAST "rpm:vendor",
+                       BAD_CAST package->rpm_vendor);
 
 
     /***********************************
      Element: group
     ************************************/
 
-    xmlNewTextChild(format, NULL,  BAD_CAST "rpm:group",
-                    BAD_CAST ((package->rpm_group) ? package->rpm_group : ""));
+    cr_xmlNewTextChild(format, NULL,  BAD_CAST "rpm:group",
+                       BAD_CAST package->rpm_group);
 
 
     /***********************************
      Element: buildhost
     ************************************/
 
-    xmlNewTextChild(format, NULL,  BAD_CAST "rpm:buildhost",
-                    BAD_CAST ((package->rpm_buildhost) ? package->rpm_buildhost : ""));
+    cr_xmlNewTextChild(format, NULL,  BAD_CAST "rpm:buildhost",
+                       BAD_CAST package->rpm_buildhost);
 
 
     /***********************************
      Element: sourcerpm
     ************************************/
 
-    xmlNewTextChild(format, NULL,  BAD_CAST "rpm:sourcerpm",
-                    BAD_CAST ((package->rpm_sourcerpm) ? package->rpm_sourcerpm : ""));
+    cr_xmlNewTextChild(format, NULL,  BAD_CAST "rpm:sourcerpm",
+                       BAD_CAST package->rpm_sourcerpm);
 
 
     /***********************************