hmmlib2: re-instsate "offset" for marker/ref + fixes

fixes:
    - indentation
    - parsing bug for marker with both episode + parameter

hmmlib2/hmmlib.h

@@ -16,14 +16,14 @@ typedef struct {
 } HMML_VideoCustomMetaData;
 
 typedef struct {
-	char* member;
+    char* member;
-	char* stream_platform;
+    char* stream_platform;
-	char* stream_username;
+    char* stream_username;
-	char* project;
+    char* project;
-	char* title;
+    char* title;
-	char* vod_platform;
+    char* vod_platform;
-	char* id;
+    char* id;
-	char* output;
+    char* output;
     char* template;
     char* medium;
 
@@ -36,67 +36,69 @@ typedef struct {
 } HMML_VideoMetaData;
 
 typedef struct {
-	char* site;
+    char* site;
-	char* page;
+    char* page;
-	char* url;
+    char* url;
-	char* title;
+    char* title;
-	char* article;
+    char* article;
-	char* author;
+    char* author;
-	char* editor;
+    char* editor;
-	char* publisher;
+    char* publisher;
-	char* isbn;
+    char* isbn;
+    int offset;
 } HMML_Reference;
 
 typedef enum {
-	HMML_CATEGORY,
+    HMML_CATEGORY,
-	HMML_MEMBER,
+    HMML_MEMBER,
-	HMML_PROJECT,
+    HMML_PROJECT,
 
-	HMML_MARKER_COUNT,
+    HMML_MARKER_COUNT,
 } HMML_MarkerType;
 
 typedef struct {
-	HMML_MarkerType type;
+    HMML_MarkerType type;
-	char* marker;
+    char* marker;
-	char* parameter;
+    char* parameter;
-	char* episode;
+    char* episode;
+    int offset;
 } HMML_Marker;
 
 typedef struct {
     _Bool present;
-	int id;
+    int id;
-	char* author;
+    char* author;
 } HMML_Quote;
 
 typedef struct {
-	int line;
+    int line;
 
     int h, m, s;
 
-	char* text;
+    char* text;
-	char* author;
+    char* author;
 
-	HMML_Reference* references;
+    HMML_Reference* references;
-	size_t reference_count;
+    size_t reference_count;
 
-	HMML_Marker* markers;
+    HMML_Marker* markers;
-	size_t marker_count;
+    size_t marker_count;
 
-	HMML_Quote quote;
+    HMML_Quote quote;
 } HMML_Annotation;
 
 typedef struct {
-	int line;
+    int line;
     int col;
-	char* message;
+    char* message;
 } HMML_Error;
 
 typedef struct {
-	_Bool              well_formed;
+    _Bool              well_formed;
-	HMML_VideoMetaData metadata;
+    HMML_VideoMetaData metadata;
-	HMML_Annotation*   annotations;
+    HMML_Annotation*   annotations;
-	size_t             annotation_count;
+    size_t             annotation_count;
-	HMML_Error         error;
+    HMML_Error         error;
     void*              free_list; // implementation detail
 } HMML_Output;
 
@@ -108,7 +110,7 @@ void        hmml_free  (HMML_Output* output);
 // Version
 
 extern const struct HMML_Version {
-	int Major, Minor, Patch;
+    int Major, Minor, Patch;
 } hmml_version;
 
 #endif
@@ -356,7 +358,9 @@ static HMML_Marker _hmml_parse_marker(struct _hmml_parser* p)
         ++p->cursor;
     }
 
-    HMML_Marker marker = {};
+    HMML_Marker marker = {
+        .offset = -1,
+    };
 
     char c = *p->cursor;
     if(c == '~') {
@@ -381,8 +385,10 @@ static HMML_Marker _hmml_parse_marker(struct _hmml_parser* p)
             ++p->cursor;
             size_t n = strcspn(p->cursor, " ");
             marker.episode = _hmml_persist_str(p, (struct _hmml_str){ p->cursor, n });
-            p->cursor += n;
+            p->cursor += n + 1;
-        } else if(*p->cursor != ']') {
+        }
+
+        if(*p->cursor != ']') {
             const char* end = p->cursor;
 
             for(;;) {
@@ -415,7 +421,9 @@ static HMML_Marker _hmml_parse_marker(struct _hmml_parser* p)
 
 static HMML_Reference _hmml_parse_ref(struct _hmml_parser* p)
 {
-    HMML_Reference ref = {};
+    HMML_Reference ref = {
+        .offset = -1,
+    };
 
     struct str_attr {
         struct _hmml_str str;
@@ -503,7 +511,7 @@ static void _hmml_parse_timecode(struct _hmml_parser* p, HMML_Annotation* anno)
     anno->s = s;
 }
 
-static void _hmml_parse_text(struct _hmml_parser* p, HMML_Annotation* anno)
+static size_t _hmml_parse_text(struct _hmml_parser* p, HMML_Annotation* anno)
 {
     static char text_mem[4096];
     char* out = text_mem;
@@ -553,9 +561,11 @@ static void _hmml_parse_text(struct _hmml_parser* p, HMML_Annotation* anno)
             if(strncmp(p->cursor + 1, "ref", 3) == 0) {
                 p->cursor += 4;
                 HMML_Reference ref = _hmml_parse_ref(p);
+                ref.offset = out - text_mem;
                 _hmml_persist_array(p, &anno->references, &anno->reference_count, ref);
             } else {
                 HMML_Marker m = _hmml_parse_marker(p);
+                m.offset = out - text_mem;
                 _hmml_persist_array(p, &anno->markers, &anno->marker_count, m);
 
                 size_t text_len = strlen(m.marker);
@@ -573,7 +583,9 @@ static void _hmml_parse_text(struct _hmml_parser* p, HMML_Annotation* anno)
                 ++p->cursor;
             } else {
                 HMML_Marker m = _hmml_parse_marker(p);
+                m.offset = out - text_mem;
                 _hmml_persist_array(p, &anno->markers, &anno->marker_count, m);
+
                 size_t text_len = strlen(m.marker);
                 CHECKSIZE(text_len);
                 memcpy(out, m.marker, text_len);
@@ -592,9 +604,12 @@ static void _hmml_parse_text(struct _hmml_parser* p, HMML_Annotation* anno)
         --out;
     }
 
-    anno->text = _hmml_persist_str(p, (struct _hmml_str){ text_mem, out - text_mem });
+    size_t text_size = out - text_mem;
+    anno->text = _hmml_persist_str(p, (struct _hmml_str){ text_mem, text_size });
 
 #undef CHECKSIZE
+
+    return text_size;
 }
 
 static void _hmml_parse_quote(struct _hmml_parser* p, HMML_Annotation* anno)
@@ -647,7 +662,7 @@ static void _hmml_parse_annotations(struct _hmml_parser* p)
 
         ++p->cursor;
 
-        _hmml_parse_text(p, &anno);
+        int text_len = _hmml_parse_text(p, &anno);
 
         if(p->cursor[0] == '[' && p->cursor[1] == ':') {
             HMML_Marker m = _hmml_parse_marker(p);
@@ -658,7 +673,7 @@ static void _hmml_parse_annotations(struct _hmml_parser* p)
             _hmml_parse_quote(p, &anno);
         }
 
-        // convert all markers to lowercase
+        // convert all markers to lowercase, fix any out of range offsets
         for(size_t i = 0; i < anno.marker_count; ++i) {
             HMML_Marker* m = anno.markers + i;
             for(char* c = m->marker; *c; ++c) {
@@ -666,6 +681,9 @@ static void _hmml_parse_annotations(struct _hmml_parser* p)
                     *c = (*c - ('A' - 'a'));
                 }
             }
+            if(m->offset > text_len) {
+                m->offset = text_len;
+            }
         }
 
         _hmml_persist_array(p, &p->out.annotations, &p->out.annotation_count, anno);