Annotation-System/hmmlib/hmmlib.l

%{
#include <stdio.h>
#include <stdbool.h>
#include <ctype.h>
#include "stb_sb.h"
#include "hmmlib.h"

	const struct HMML_Version hmml_version = { 0, 1, 0 };

	typedef struct {
		int    line;
		HMML_Annotation* annos;

		HMML_VideoMetaData meta;
		HMML_Annotation an;
		HMML_Reference  ref;

		HMML_Error error;

		void*  attr;
		int    mnext;
		bool   first;
	} HMML_ParseState;

#define HMML_ERR(fmt, ...) \
	do { \
		asprintf(&yyextra->error.message, fmt, ##__VA_ARGS__);\
		yyextra->error.line = yyextra->line;\
		return 1;\
	} while(0)

#define CHECKESCAPE(x) do { if(!strchr("[]:@~\\\"", x)) HMML_ERR("Unknown backslash escape code '%c'", x); } while(0)

#define V_(x) &yyextra->meta.x
#define R_(x) &yyextra->ref.x
#define M_(x, state) do { HMML_Marker m = { HMML_ ## x }; sb_push(yyextra->an.markers, m); yyextra->mnext = state; } while(0)
#define M_ADD(t, n) \
	do { \
		char* c = UNQUOTE(strndup(t, n));\
		sb_last(yyextra->an.markers).marker = c;\
		sb_last(yyextra->an.markers).offset = sb_count(yyextra->an.text);\
		if(yyextra->mnext == TEXT){\
			memcpy(sb_add(yyextra->an.text, n), c, n);\
		}\
	} while(0)

#define MX_ADD(c) \
	do { \
		sb_push(sb_last(yyextra->an.markers).parameter, c);\
		if(c) sb_push(yyextra->an.text, c);\
	} while(0)

#define NEWANNO() \
	do { \
		if(!yyextra->first) sb_push(yyextra->annos, yyextra->an); \
		memset(&yyextra->an, 0, sizeof(yyextra->an));\
		yyextra->an.line = yyextra->line;\
		yyextra->first = false;\
	} while(0)

#define UNQUOTE(_attr) ({ \
	typeof(_attr) attr = (_attr); \
	size_t len = strlen(attr); \
	for(char* c = attr; *c; ++c){\
		if(*c == '\\'){ \
			CHECKESCAPE(c[1]); \
			memmove(c, c+1, len-(c-attr)); \
			--len; \
		} \
	} \
	attr; \
})

%}

%option reentrant
%option extra-type="HMML_ParseState*"
%option noyywrap

S               [\t \r]*
ATTR_SIMPLE     [^\" \]\t\r\n][^ \]\t\r\n]*
ATTR_ALNUM      [0-9a-zA-Z][0-9a-zA-Z_]*
ATTR_QUOTED     \"([^\n\"\\]|\\.)*\"
TAG_VIDEO_OPEN  \[video
TIMECODE        \[[0-9]{1,2}(:[0-5][0-9]){1,2}\]
BAD_TIMECODE    \[[0-9]{1,2}(:[6-9][0-9]){1,2}\]
TEXT_BREAK      [^\\:@~\[\]\r\n\t ]
LB              \[
RB              \]

%s VIDEO
%s V_ATTR
%s V2_ATTR
%s ANNOTATION
%s TEXT_START
%s TEXT
%s MARKER
%s MARKER_XTRA
%s REF
%s R_ATTR
%s AFTERTEXT
%s AUTHOR
%s CATEGORIES
%s QUOTES

%%

<<EOF>> { HMML_ERR("Unexpected EOF, video close tag not found."); }
\r\n|\n { yyextra->line++; }

<INITIAL>{TAG_VIDEO_OPEN} { BEGIN(VIDEO); }
<INITIAL>.                { HMML_ERR("Missing video tag."); }

<VIDEO>{S}
<VIDEO>member{S}=          { yyextra->attr = V_(member);          BEGIN(V_ATTR); }
<VIDEO>stream_platform{S}= { yyextra->attr = V_(stream_platform); BEGIN(V_ATTR); }
<VIDEO>stream_username{S}= { yyextra->attr = V_(stream_username); BEGIN(V_ATTR); }
<VIDEO>project{S}=         { yyextra->attr = V_(project);         BEGIN(V_ATTR); }
<VIDEO>title{S}=           { yyextra->attr = V_(title);           BEGIN(V_ATTR); }
<VIDEO>vod_platform{S}=    { yyextra->attr = V_(vod_platform);    BEGIN(V_ATTR); }
<VIDEO>id{S}=              { yyextra->attr = V_(id);              BEGIN(V_ATTR); }
<VIDEO>co\-host{S}=        { yyextra->attr = V_(co_hosts);        BEGIN(V2_ATTR); }
<VIDEO>guest{S}=           { yyextra->attr = V_(guests);          BEGIN(V2_ATTR); }
<VIDEO>annotator{S}=       { yyextra->attr = V_(annotators);      BEGIN(V2_ATTR); }
<VIDEO>\]                  { BEGIN(ANNOTATION); };
<VIDEO>.                   { HMML_ERR("Invalid char '%s' in video tag.", yytext); }

<V_ATTR>{S}           { BEGIN(VIDEO); }
<V_ATTR>{ATTR_SIMPLE} { *(char**)yyextra->attr = strndup(yytext  , yyleng  ); BEGIN(VIDEO); }
<V_ATTR>{ATTR_QUOTED} { *(char**)yyextra->attr = UNQUOTE(strndup(yytext+1, yyleng-2)); BEGIN(VIDEO); }
<V_ATTR>\]            { yyless(0); BEGIN(VIDEO); }

<V2_ATTR>{S}           { BEGIN(VIDEO); }
<V2_ATTR>{ATTR_SIMPLE} { sb_push(*(char***)yyextra->attr, strndup(yytext  , yyleng  )); BEGIN(VIDEO); }
<V2_ATTR>{ATTR_QUOTED} { sb_push(*(char***)yyextra->attr, UNQUOTE(strndup(yytext+1, yyleng-2))); BEGIN(VIDEO); }
<V2_ATTR>\]            { yyless(0); BEGIN(VIDEO); }

<ANNOTATION>{TIMECODE}{LB}@ { NEWANNO(); yyextra->an.time = strndup(yytext+1, yyleng-4); BEGIN(AUTHOR); }
<ANNOTATION>{TIMECODE}      { NEWANNO(); yyextra->an.time = strndup(yytext+1, yyleng-2); BEGIN(TEXT_START); }
<ANNOTATION>{BAD_TIMECODE}  { HMML_ERR("Timecode %s out of range.", yytext); }
<ANNOTATION>{S}
<ANNOTATION>.               { HMML_ERR("Cannot parse annotation. Expected timecode."); }

<TEXT_START>{LB}: { M_(CATEGORY, TEXT); BEGIN(MARKER); }
<TEXT_START>{LB}@ { M_(MEMBER  , TEXT); BEGIN(MARKER); }
<TEXT_START>{LB}~ { M_(PROJECT , TEXT); BEGIN(MARKER); }
<TEXT_START>{LB}  { BEGIN(TEXT); }
<TEXT_START>.     { HMML_ERR("Unknown character '%s' after timecode.", yytext); }

<TEXT>{TEXT_BREAK}+ { memcpy(sb_add(yyextra->an.text, yyleng), yytext, yyleng); }
<TEXT>\\.           { CHECKESCAPE(yytext[1]); memcpy(sb_add(yyextra->an.text, yyleng-1), yytext+1, yyleng-1); }
<TEXT>[ \r\t]+:     { sb_push(yyextra->an.text, ' '); M_(CATEGORY, TEXT); BEGIN(MARKER); }
<TEXT>[ \r\t]+@     { sb_push(yyextra->an.text, ' '); M_(MEMBER  , TEXT); BEGIN(MARKER); }
<TEXT>[ \r\t]+~     { sb_push(yyextra->an.text, ' '); M_(PROJECT , TEXT); BEGIN(MARKER); }
<TEXT>{LB}:         { M_(CATEGORY, MARKER_XTRA); BEGIN(MARKER); }
<TEXT>{LB}@         { M_(MEMBER  , MARKER_XTRA); BEGIN(MARKER); }
<TEXT>{LB}~         { M_(PROJECT , MARKER_XTRA); BEGIN(MARKER); }
<TEXT>\]            { BEGIN(AFTERTEXT); }
<TEXT>{LB}ref       { yyextra->ref.offset = sb_count(yyextra->an.text); BEGIN(REF); }
<TEXT>{LB}
<TEXT>{S}           { sb_push(yyextra->an.text, ' '); }
<TEXT>.             { sb_push(yyextra->an.text, *yytext); }

<MARKER>{ATTR_ALNUM}  { M_ADD(yytext  , yyleng  ); BEGIN(yyextra->mnext); };
<MARKER>{ATTR_QUOTED} { M_ADD(yytext+1, yyleng-2); BEGIN(yyextra->mnext); };
<MARKER>.             { HMML_ERR("Cannot parse Marker. Expected quoted or alphanumeric attribute."); }

<MARKER_XTRA>\\]   { MX_ADD(']'); }
<MARKER_XTRA>\\#   { MX_ADD('#'); }
<MARKER_XTRA>\]    { MX_ADD('\0'); BEGIN(TEXT); }
<MARKER_XTRA>[ ]\#[0-9]+ {
	HMML_Marker* m = &sb_last(yyextra->an.markers);
	if(m->type == HMML_PROJECT){
		m->episode = strdup(yytext+2);
	} else {
		memcpy(sb_add(m->parameter, yyleng-1), yytext+1, yyleng-1);
		memcpy(sb_add(yyextra->an.text, yyleng-1), yytext+1, yyleng-1);
	}
}
<MARKER_XTRA>[ ]   { if(sb_last(yyextra->an.markers).parameter){ MX_ADD(' '); } }
<MARKER_XTRA>.     { MX_ADD(*yytext); }

<REF>{S}
<REF>site{S}=      { yyextra->attr = R_(site);      BEGIN(R_ATTR); }
<REF>page{S}=      { yyextra->attr = R_(page);      BEGIN(R_ATTR); }
<REF>url{S}=       { yyextra->attr = R_(url);       BEGIN(R_ATTR); }
<REF>title{S}=     { yyextra->attr = R_(title);     BEGIN(R_ATTR); }
<REF>article{S}=   { yyextra->attr = R_(article);   BEGIN(R_ATTR); }
<REF>author{S}=    { yyextra->attr = R_(author);    BEGIN(R_ATTR); }
<REF>editor{S}=    { yyextra->attr = R_(editor);    BEGIN(R_ATTR); }
<REF>publisher{S}= { yyextra->attr = R_(publisher); BEGIN(R_ATTR); }
<REF>isbn{S}=      { yyextra->attr = R_(isbn);      BEGIN(R_ATTR); }
<REF>\]            { sb_push(yyextra->an.references, yyextra->ref); memset(&yyextra->ref, 0, sizeof(yyextra->ref)); BEGIN(TEXT); }
<REF>.             { HMML_ERR("Unexpected item in ref: %s", yytext); }

<R_ATTR>{S}
<R_ATTR>{ATTR_SIMPLE} { *(char**)yyextra->attr = strndup(yytext, yyleng); BEGIN(REF); }
<R_ATTR>{ATTR_QUOTED} { *(char**)yyextra->attr = UNQUOTE(strndup(yytext+1, yyleng-2)); BEGIN(REF); }

<AFTERTEXT,ANNOTATION>\[\/video\] { NEWANNO(); return 0; }

<AFTERTEXT>{S}
<AFTERTEXT>{LB}quote   { BEGIN(QUOTES); }
<AFTERTEXT>{LB}:       { BEGIN(CATEGORIES); yyless(1); }
<AFTERTEXT>{LB}[0-9]   { BEGIN(ANNOTATION); yyless(0); }
<AFTERTEXT>..          { HMML_ERR("Unexpected thing after text node: %s", yytext); }
<AFTERTEXT>.           { HMML_ERR("Unexpected thing after text node: %s", yytext); }

<AUTHOR>[^\]\n]+\] { yyextra->an.author = strndup(yytext, yyleng-1); BEGIN(TEXT_START); }
<AUTHOR>{S}

<CATEGORIES>{S} 
<CATEGORIES>:{ATTR_SIMPLE} { HMML_Marker m = { HMML_CATEGORY, strndup(yytext+1, yyleng-1), NULL, NULL, -1 }; sb_push(yyextra->an.markers, m); }
<CATEGORIES>:{ATTR_QUOTED} { HMML_Marker m = { HMML_CATEGORY, UNQUOTE(strndup(yytext+2, yyleng-3)), NULL, NULL, -1 }; sb_push(yyextra->an.markers, m); }
<CATEGORIES>\]{LB}quote    { BEGIN(QUOTES); }
<CATEGORIES>\]             { BEGIN(ANNOTATION); }
<CATEGORIES>.              { HMML_ERR("Unexpected character in category tag: '%s'\n", yytext); }

<QUOTES>{S}
<QUOTES>[0-9]+{S}\]  { yyextra->an.is_quote = true; yyextra->an.quote.id = atoi(yytext); BEGIN(ANNOTATION); }
<QUOTES>{ATTR_ALNUM} { memcpy(sb_add(yyextra->an.quote.author, yyleng), yytext, yyleng); }
<QUOTES>.            { HMML_ERR("Unexpected character in quotes tag: %s", yytext); }

%%

#define HMML_REF_ITEMS 9

static void _hmml_free_ref(HMML_Reference*);
static void _hmml_free_anno(HMML_Annotation*);

HMML_Output hmml_parse_file(FILE* f){
	HMML_Output     output = {};
	HMML_ParseState state  = {};

	state.first = true;
	state.line = 1;

	yyscan_t scan;
	yylex_init_extra(&state, &scan);
	yyset_in(f, scan);

	output.well_formed = yylex(scan) == 0;

	memcpy(&output.metadata, &state.meta, sizeof(HMML_VideoMetaData));

	output.metadata.co_host_count   = sb_count(output.metadata.co_hosts);
	output.metadata.guest_count     = sb_count(output.metadata.guests);
	output.metadata.annotator_count = sb_count(output.metadata.annotators);

	output.annotations = state.annos;
	output.annotation_count = sb_count(state.annos);

	for(size_t i = 0; i < output.annotation_count; ++i){
		HMML_Annotation* a = output.annotations + i;
		sb_push(a->text, 0);
		a->marker_count = sb_count(a->markers);
		a->reference_count = sb_count(a->references);

		for(size_t j = 0; j < sb_count(a->markers); ++j){
			for(char* c = a->markers[j].marker; *c; ++c) *c = tolower(*c);
		}

		if(a->is_quote && a->quote.author){
			sb_push(a->quote.author, 0);
		}
	}

	if(!output.well_formed){
		hmml_free(&output);
		memcpy(&output.error, &state.error, sizeof(HMML_Error));
	}

	_hmml_free_anno(&state.an);
	_hmml_free_ref(&state.ref);

	yylex_destroy(scan);

	return output;
}

static void _hmml_free_ref(HMML_Reference* r){
	for(size_t k = 0; k < HMML_REF_ITEMS; ++k){
		free(((char**)r)[k]);
	}
}

static void _hmml_free_anno(HMML_Annotation* a){
	free(a->time);
	free(a->author);
	sb_free(a->text);

	sb_each(r, a->references){
		_hmml_free_ref(r);
	}
	sb_free(a->references);

	sb_each(m, a->markers){
		free(m->marker);
		free(m->episode);
		sb_free(m->parameter);
	}
	sb_free(a->markers);

	sb_free(a->quote.author);
}

void hmml_free(HMML_Output* hmml){
	if(!hmml) return;

	for(size_t i = 0; i < hmml->annotation_count; ++i){
		_hmml_free_anno(hmml->annotations + i);
	}

	sb_free(hmml->annotations);

	free(hmml->metadata.member);
	free(hmml->metadata.stream_platform);
	free(hmml->metadata.stream_username);
	free(hmml->metadata.project);
	free(hmml->metadata.title);
	free(hmml->metadata.vod_platform);
	free(hmml->metadata.id);

	sb_each(i, hmml->metadata.co_hosts)   free(*i);
	sb_each(i, hmml->metadata.guests)     free(*i);
	sb_each(i, hmml->metadata.annotators) free(*i);

	sb_free(hmml->metadata.co_hosts);
	sb_free(hmml->metadata.guests);
	sb_free(hmml->metadata.annotators);

	free(hmml->error.message);
	memset(hmml, 0, sizeof(*hmml));
}

typedef struct {
	char* text;
	int*  lines;
} Index;

Index* index_find(Index* base, const char* text){
	for(size_t i = 0; i < sb_count(base); ++i){
		if(strcmp(base[i].text, text) == 0){
			return base + i;
		}
	}
	return NULL;
}

void hmml_dump(HMML_Output* hmml){

	if(!hmml){
		puts("(null");
		return;
	}

	if(!hmml->well_formed){
		printf("Parse error on line %d: %s\n", hmml->error.line, hmml->error.message);
		return;
	}


	puts("Annotations:");
	for(size_t i = 0; i < hmml->annotation_count; ++i){
		HMML_Annotation* a = hmml->annotations + i;
		printf("\t%3d [%7s] [%s]\n", a->line, a->time, a->text);
	}

	Index* authors = NULL;
	Index* markers[HMML_MARKER_COUNT] = {};
	int max_text_len = 0;

	for(size_t i = 0; i < hmml->annotation_count; ++i){
		HMML_Annotation* a = hmml->annotations + i;

		if(a->author){
			size_t len = strlen(a->author);
			if(len > max_text_len){
				max_text_len = len;
			}

			Index* idx;
			if(!(idx = index_find(authors, a->author))){
				Index x = { a->author };
				sb_push(authors, x);
				idx = &sb_last(authors);
			}

			sb_push(idx->lines, a->line);
		}
	}

	for(size_t i = 0; i < hmml->annotation_count; ++i){
		HMML_Annotation* a = hmml->annotations + i;

		for(size_t j = 0; j < a->marker_count; ++j){
			int   type = a->markers[j].type;
			char* text = a->markers[j].marker;

			size_t len = strlen(text);
			if(len > max_text_len){
				max_text_len = len;
			}

			Index* idx;
			if(!(idx = index_find(markers[type], text))){
				Index x = { text };
				sb_push(markers[type], x);
				idx = &sb_last(markers[type]);
			}

			sb_push(idx->lines, a->line);
		}
	}

	puts("Authors:");
	for(size_t i = 0; i < sb_count(authors); ++i){
		printf("\t %*s: ", max_text_len, authors[i].text);
		for(size_t j = 0; j < sb_count(authors[i].lines); ++j){
			printf("%3d ", authors[i].lines[j]);
		}
		puts("");
	}


	static const char* m_tags[HMML_MARKER_COUNT] = { "Categories", "Members", "Projects" };

	for(size_t i = 0; i < HMML_MARKER_COUNT; ++i){
		printf("%s:\n", m_tags[i]);
		for(size_t j = 0; j < sb_count(markers[i]); ++j){
			printf("\t %*s: ", max_text_len, markers[i][j].text);
			for(size_t k = 0; k < sb_count(markers[i][j].lines); ++k){
				printf("%3d ", markers[i][j].lines[k]);
			}
			puts("");
		}
	}

	static const char* r_tags[] = { "Site", "Page", "URL", "Title", "Article", "Author", "Editor", "Publisher", "ISBN" };
	puts("References:");
	for(size_t i = 0; i < hmml->annotation_count; ++i){
		HMML_Annotation* a = hmml->annotations + i;
		for(size_t j = 0; j < a->reference_count; ++j){
			printf("\t%3d ", a->line);
			HMML_Reference* r = a->references + j;
			for(size_t k = 0; k < HMML_REF_ITEMS; ++k){
				char* item = ((char**)r)[k];
				if(item){
					printf("[%s = %s] ", r_tags[k], item);
				}
			}
			puts("");
		}
	}

	puts("Quotes:");
	for(size_t i = 0; i < hmml->annotation_count; ++i){
		HMML_Annotation* a = hmml->annotations + i;
		if(a->is_quote){
			if(a->quote.author){
				printf("\t%3d [Quote #%d, by %s]", a->line, a->quote.id, a->quote.author);
			} else {
				printf("\t%3d [Quote #%d]", a->line, a->quote.id);
			}
			puts("");
		}
	}

	for(size_t i = 0; i < sb_count(authors); ++i){
		sb_free(authors[i].lines);
	}
	sb_free(authors);

	for(size_t i = 0; i < HMML_MARKER_COUNT; ++i){
		for(size_t j = 0; j < sb_count(markers[i]); ++j){
			sb_free(markers[i][j].lines);
		}
		sb_free(markers[i]);
	}
}