dvander@145: #include <stdio.h>
dvander@145: #include <ctype.h>
dvander@150: #include <wctype.h>
dvander@145: #include <string.h>
dvander@150: #include <stdlib.h>
dvander@145: #include "CTextParsers.h"
dvander@145: 
dvander@145: CTextParsers g_TextParse;
dvander@145: 
dvander@145: static int g_ini_chartable1[255] = {0};
dvander@150: static int g_ws_chartable[255] = {0};
dvander@145: 
dvander@145: CTextParsers::CTextParsers()
dvander@145: {
dvander@145: 	g_ini_chartable1['_'] = 1;
dvander@145: 	g_ini_chartable1['-'] = 1;
dvander@145: 	g_ini_chartable1[','] = 1;
dvander@145: 	g_ini_chartable1['+'] = 1;
dvander@145: 	g_ini_chartable1['.'] = 1;
dvander@145: 	g_ini_chartable1['$'] = 1;
dvander@145: 	g_ini_chartable1['?'] = 1;
dvander@145: 	g_ini_chartable1['/'] = 1;
dvander@150: 	g_ws_chartable['\n'] = 1;
dvander@150: 	g_ws_chartable['\v'] = 1;
dvander@150: 	g_ws_chartable['\r'] = 1;
dvander@150: 	g_ws_chartable['\t'] = 1;
dvander@150: 	g_ws_chartable['\f'] = 1;
dvander@150: 	g_ws_chartable[' '] = 1;
dvander@150: }
dvander@150: 
dvander@150: unsigned int CTextParsers::GetUTF8CharBytes(const char *stream)
dvander@150: {
dvander@150: 	return _GetUTF8CharBytes(stream);
dvander@145: }
dvander@145: 
dvander@159: bool CTextParsers::ParseFile_SMC(const char *file, ITextListener_SMC *smc_listener, unsigned int *line, unsigned int *col, bool strict)
dvander@145: {
dvander@159: 	/* This beast is a lot more rigorous than our INI friend. */
dvander@145: 
dvander@145: 	return false;
dvander@145: }
dvander@145: 
dvander@145: bool CTextParsers::ParseFile_INI(const char *file, ITextListener_INI *ini_listener, unsigned int *line, unsigned int *col)
dvander@145: {
dvander@145: 	FILE *fp = fopen(file, "rt");
dvander@145: 	unsigned int curline = 0;
dvander@145: 	unsigned int curtok;
dvander@145: 	size_t len;
dvander@145: 
dvander@145: 	if (!fp)
dvander@145: 	{
dvander@145: 		if (line)
dvander@145: 		{
dvander@145: 			*line = 0;
dvander@145: 		}
dvander@145: 
dvander@145: 		return false;
dvander@145: 	}
dvander@145: 
dvander@145: 	char buffer[2048];
dvander@145: 	char *ptr, *save_ptr;
dvander@145: 	bool in_quote;
dvander@150: 
dvander@145: 	while (!feof(fp))
dvander@145: 	{
dvander@145: 		curline++;
dvander@145: 		curtok = 0;
dvander@145: 		buffer[0] = '\0';
dvander@145: 		if (fgets(buffer, sizeof(buffer), fp) == NULL)
dvander@145: 		{
dvander@145: 			break;
dvander@145: 		}
dvander@150: 
dvander@150: 		//:TODO: this will only run once, so find a nice way to move it out of the while loop
dvander@150: 		/* If this is the first line, check the first three bytes for BOM */
dvander@150: 		if (curline == 1 && 
dvander@150: 			buffer[0] == (char)0xEF && 
dvander@150: 			buffer[1] == (char)0xBB && 
dvander@150: 			buffer[2] == (char)0xBF)
dvander@150: 		{
dvander@150: 			/* We have a UTF-8 marked file... skip these bytes */
dvander@150: 			ptr = &buffer[3];
dvander@150: 		} else {
dvander@150: 			ptr = buffer;
dvander@150: 		}
dvander@150: 
dvander@150: 		/***************************************************
dvander@150: 		 * We preprocess the string before parsing tokens! *
dvander@150: 		 ***************************************************/
dvander@150: 
dvander@145: 		/* First strip beginning whitespace */
dvander@150: 		while (*ptr != '\0' && g_ws_chartable[*ptr] != 0)
dvander@145: 		{
dvander@145: 			ptr++;
dvander@145: 		}
dvander@145: 
dvander@145: 		len = strlen(ptr);
dvander@145: 
dvander@145: 		if (!len)
dvander@145: 		{
dvander@145: 			continue;
dvander@145: 		}
dvander@145: 
dvander@145: 		/* Now search for comment characters */
dvander@145: 		in_quote = false;
dvander@145: 		save_ptr = ptr;
dvander@145: 		for (size_t i=0; i<len; i++,ptr++)
dvander@145: 		{
dvander@145: 			if (!in_quote)
dvander@145: 			{
dvander@145: 				switch (*ptr)
dvander@145: 				{
dvander@145: 				case '"':
dvander@145: 					{
dvander@145: 						in_quote = true;
dvander@145: 						break;
dvander@145: 					}
dvander@145: 				case ';':
dvander@145: 					{
dvander@145: 						/* Stop the loop */
dvander@145: 						len = i;
dvander@145: 						/* Terminate the string here */
dvander@145: 						*ptr = '\0';
dvander@145: 						break;
dvander@145: 					}
dvander@145: 				}
dvander@145: 			} else {
dvander@145: 				if (*ptr == '"')
dvander@145: 				{
dvander@145: 					in_quote = false;
dvander@145: 				}
dvander@145: 			}
dvander@145: 		}
dvander@145: 
dvander@145: 		if (!len)
dvander@145: 		{
dvander@145: 			continue;
dvander@145: 		}
dvander@145: 
dvander@145: 		ptr = save_ptr;
dvander@145: 
dvander@145: 		/* Lastly, strip ending whitespace off */
dvander@145: 		for (size_t i=len-1; i>=0 && i<len; i--)
dvander@145: 		{
dvander@150: 			if (g_ws_chartable[ptr[i]])
dvander@145: 			{
dvander@145: 				ptr[i] = '\0';
dvander@145: 				len--;
dvander@145: 			} else {
dvander@145: 				break;
dvander@145: 			}
dvander@145: 		}
dvander@145: 
dvander@145: 		if (!len)
dvander@145: 		{
dvander@145: 			continue;
dvander@145: 		}
dvander@145: 
dvander@145: 		if (!ini_listener->ReadINI_RawLine(ptr, &curtok))
dvander@145: 		{
dvander@145: 			goto event_failed;
dvander@145: 		}
dvander@145: 
dvander@145: 		if (*ptr == '[')
dvander@145: 		{
dvander@145: 			bool invalid_tokens = false;
dvander@145: 			bool got_bracket = false;
dvander@145: 			bool extra_tokens = false;
dvander@145: 			char c;
dvander@150: 			bool alnum;
dvander@150: 			wchar_t wc;
dvander@145: 
dvander@145: 			for (size_t i=1; i<len; i++)
dvander@145: 			{
dvander@145: 				c = ptr[i];
dvander@150: 				alnum = false;
dvander@150: 
dvander@150: 				if (c & (1<<7))
dvander@150: 				{
dvander@150: 					if (mbtowc(&wc, &ptr[i], len-i) != -1)
dvander@150: 					{
dvander@150: 						alnum = (iswalnum(wc) != 0);
dvander@150: 						i += _GetUTF8CharBytes(&ptr[i]) - 1;
dvander@150: 					}
dvander@150: 				} else {
dvander@150: 					alnum = (isalnum(c) != 0) || (g_ini_chartable1[c] != 0);
dvander@150: 				}
dvander@150: 				if (!alnum)
dvander@145: 				{
dvander@145: 					/* First check - is this a bracket? */
dvander@145: 					if (c == ']')
dvander@145: 					{
dvander@145: 						/* Yes! */
dvander@145: 						got_bracket = true;
dvander@145: 						/* If this isn't the last character... */
dvander@145: 						if (i != len - 1)
dvander@145: 						{
dvander@145: 							extra_tokens = true;
dvander@145: 						}
dvander@145: 						/* terminate */
dvander@145: 						ptr[i] = '\0';
dvander@145: 						break;
dvander@145: 					} else {
dvander@145: 						/* n...No! Continue copying. */
dvander@145: 						invalid_tokens = true;
dvander@145: 					}
dvander@145: 				}
dvander@145: 			}
dvander@145: 
dvander@145: 			/* Tell the handler */
dvander@145: 			if (!ini_listener->ReadINI_NewSection(&ptr[1], invalid_tokens, got_bracket, extra_tokens, &curtok))
dvander@145: 			{
dvander@145: 				goto event_failed;
dvander@145: 			}
dvander@145: 		} else {
dvander@145: 			char *key_ptr = ptr;
dvander@145: 			char *val_ptr = NULL;
dvander@145: 			char c;
dvander@145: 			size_t first_space = 0;
dvander@145: 			bool invalid_tokens = false;
dvander@145: 			bool equal_token = false;
dvander@145: 			bool quotes = false;
dvander@150: 			bool alnum;
dvander@150: 			wchar_t wc;
dvander@150: 
dvander@145: 			for (size_t i=0; i<len; i++)
dvander@145: 			{
dvander@145: 				c = ptr[i];
dvander@150: 				alnum = false;
dvander@145: 				/* is this an invalid char? */
dvander@150: 				if (c & (1<<7))
dvander@145: 				{
dvander@150: 					if (mbtowc(&wc, &ptr[i], len-i) != -1)
dvander@150: 					{
dvander@150: 						alnum = (iswalnum(wc) != 0);
dvander@150: 						i += _GetUTF8CharBytes(&ptr[i]) - 1;
dvander@150: 					}
dvander@150: 				} else {
dvander@150: 					alnum = (isalnum(c) != 0) || (g_ini_chartable1[c] != 0);
dvander@150: 				}
dvander@150: 
dvander@150: 				if (!alnum)
dvander@150: 				{
dvander@150: 					if (g_ws_chartable[c])
dvander@145: 					{
dvander@159: 						/* if it's a space, keep track of the first occurring space */
dvander@145: 						if (!first_space)
dvander@145: 						{
dvander@145: 							first_space = i;
dvander@145: 						}
dvander@145: 					} else {
dvander@145: 						if (c == '=')
dvander@145: 						{
dvander@145: 							/* if it's an equal sign, we're done with the key */
dvander@145: 							if (first_space)
dvander@145: 							{
dvander@145: 								/* remove excess whitespace */
dvander@145: 								key_ptr[first_space] = '\0';
dvander@145: 							} else {
dvander@145: 								/* remove the equal sign */
dvander@145: 								key_ptr[i] = '\0';
dvander@145: 							}
dvander@145: 							if (ptr[++i] != '\0')
dvander@145: 							{
dvander@145: 								/* If this isn't the end, set next pointer */
dvander@145: 								val_ptr = &ptr[i];
dvander@145: 							}
dvander@145: 							equal_token = true;
dvander@145: 							break;
dvander@145: 						} else {
dvander@145: 							/* Mark that we got something invalid! */
dvander@145: 							invalid_tokens = true;
dvander@145: 							first_space = 0;
dvander@145: 						}
dvander@145: 					}
dvander@145: 				}
dvander@145: 			}
dvander@145: 
dvander@145: 			/* Now we need to parse the value, if any */
dvander@145: 			if (val_ptr)
dvander@145: 			{
dvander@145: 				/* eat up spaces! there shouldn't be any h*/
dvander@150: 				while ((*val_ptr != '\0') && g_ws_chartable[*val_ptr] != 0)
dvander@145: 				{
dvander@145: 					val_ptr++;
dvander@145: 				}
dvander@145: 				if (*val_ptr == '\0')
dvander@145: 				{
dvander@145: 					val_ptr = NULL;
dvander@145: 					goto skip_value;
dvander@145: 				}
dvander@145: 				/* Do we have an initial quote? If so, the parsing rules change! */
dvander@145: 				if (*val_ptr == '"' && *val_ptr != '\0')
dvander@145: 				{
dvander@145: 					len = strlen(val_ptr);
dvander@145: 					if (val_ptr[len-1] == '"')
dvander@145: 					{
dvander@145: 						/* Strip quotes! */
dvander@145: 						val_ptr[--len] = '\0';
dvander@145: 						val_ptr++;
dvander@145: 						quotes = true;
dvander@145: 					}
dvander@145: 				}
dvander@145: 			}
dvander@145: skip_value:
dvander@145: 			/* We're done! */
dvander@145: 			curtok = val_ptr - buffer;
dvander@145: 			if (!ini_listener->ReadINI_KeyValue(key_ptr, val_ptr, invalid_tokens, equal_token, quotes, &curtok))
dvander@145: 			{
dvander@145: 				curtok = 0;
dvander@145: 				goto event_failed;
dvander@145: 			}
dvander@145: 		}
dvander@145: 	}
dvander@145: 
dvander@145: 	if (line)
dvander@145: 	{
dvander@145: 		*line = curline;
dvander@145: 	}
dvander@145: 
dvander@145: 	fclose(fp);
dvander@145: 
dvander@145: 	return true;
dvander@145: 
dvander@145: event_failed:
dvander@145: 	if (line)
dvander@145: 	{
dvander@145: 		*line = curline;
dvander@145: 	}
dvander@145: 
dvander@145: 	if (col)
dvander@145: 	{
dvander@145: 		*col = curtok;
dvander@145: 	}
dvander@145: 
dvander@145: 	fclose(fp);
dvander@145: 
dvander@145: 	return false;
dvander@145: }