From: teodor <teodor>
Date: Wed, 22 Sep 2004 11:59:02 +0000 (+0000)
Subject: Initial revision
X-Git-Tag: start~1
X-Git-Url: http://sigaev.ru/git/gitweb.cgi?a=commitdiff_plain;h=d13f66d9af94b45a66a9f46c7a460b862325d858;p=clrlibru.git

Initial revision
---

d13f66d9af94b45a66a9f46c7a460b862325d858
diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..bf32e85
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,18 @@
+CC=gcc
+INCLUDE=-I.
+CFLAGS=-Wall -O2 -ansi -pedantic
+OBJS=clrlibru.o
+
+.SUFFIXES: .o.c
+
+all: clrlibru
+
+.c.o:
+	$(CC) $(CFLAGS) $(INCLUDE) -c $<
+
+clrlibru: $(OBJS)
+	$(CC) $(LIB) -o $@ $(OBJS)
+
+clean:
+	rm -rf clrlibru *core *.o
+
diff --git a/clrlibru.c b/clrlibru.c
new file mode 100644
index 0000000..13b25b6
--- /dev/null
+++ b/clrlibru.c
@@ -0,0 +1,208 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <ctype.h>
+
+static void
+usage() {
+        printf("Clear text from lib.ru for Pocket PC, Version 0.2.\n");
+        printf("Author: Teodor Sigaev <teodor@sigaev.ru>\n");
+        printf("Usage:\n   clrlibru [-i INPUTFILE] [-o OUTPUT] [-l NUMSPACE]\n");
+        exit(0);
+}
+
+char* RemoveTag[]={
+	"select",
+	"head",
+	"div",
+	"a",
+	"form",
+	NULL
+};
+
+static int
+is_rtag(char *tag, int len) {
+	char **ptr=RemoveTag;
+
+	tag[len]='\0';
+	while( *ptr ) {
+		if ( strcmp(tag, *ptr) == 0 ) 
+			return 1;
+		ptr++;
+	} 
+	return 0;
+}
+
+
+static char *optarg = NULL;
+static int current=1;
+
+int
+mgetopt(int argn, char* argv[], char *option) {
+	char key;
+
+	if ( current >= argn ) return -1;
+
+	key = *(argv[current]+1);
+	if ( *(argv[current]) == '-' || *(argv[current]) == '/' ) {
+		char *ptr = strchr( option, key );
+		if ( ptr == NULL ) {
+			printf( "Unknown option: %s\n", argv[current]);
+			exit(1);
+		}
+		current++;
+		if ( *(ptr+1) == ':' ) {
+			if (current < argn) {
+				optarg=argv[current];
+				current++;
+			} else {
+				printf("No value for -%c\n" ,key);
+				exit(1);
+			}
+		} else
+			optarg=NULL;
+	} else {
+		printf("Unknown option: %s\n", argv[current]);
+		exit(1);
+	}
+	return (int)key;
+}
+
+#define INTXT	0
+#define INTAG	1
+#define FINDEND	2
+#define INDROPTAG	3
+#define INDROPINTAG	4
+#define INHEADTAG	5
+#define INDROPINCLSTAG	6
+
+
+int 
+main(int argn, char *argv[]) {
+	int ch;
+	FILE	*in=stdin, *out=stdout;
+	int state=INTXT;
+	char	buf[8192];
+	int lenbuf=0,closelen=0;
+        int spacelen=4;
+
+	while((ch = mgetopt(argn, argv, "l:i:o:h?"))!=-1) {
+		switch (ch) {
+			case 'i':
+				if ( (in=fopen(optarg, "r"))==NULL) {
+					printf("Can't open file %s\n", optarg);
+					exit(1);
+				}
+				break;
+			case 'o':
+				if ( (out=fopen(optarg, "w"))==NULL) {
+					printf("Can't open file %s\n",optarg);
+					exit(1);
+				}
+				break;
+			case 'l':
+				spacelen = atoi(optarg);
+				if ( spacelen < 0 ) {
+					printf("-l should be >= 0\n");
+					exit(1);
+				}
+				break;
+			case 'h':
+			case '?':
+			default:
+				usage();
+		}
+	}
+	
+	while( (ch=getc(in)) != EOF ) {
+		if ( state==INTXT ) {
+			if ( ch == '<' ) {
+				state=INHEADTAG;
+				lenbuf=0;
+			} else if ( ch == '\n' ) {
+				state=FINDEND;
+				lenbuf=1;
+				*buf = ch;
+			} else if ( ch != '\r' )
+				fputc(ch,out);
+		} else if ( state==INHEADTAG ) {
+			if ( isalpha(ch) ) {
+				buf[ lenbuf ] = tolower(ch);
+				lenbuf++;
+			} else if ( ch == '>' ) {
+				if ( is_rtag(buf,lenbuf) ) {
+					state = INDROPTAG;
+					closelen=0;
+				} else {
+					state=INTXT;
+					fputc(' ',out);
+				}
+			} else if ( lenbuf == 0 && ch != '/' ) {
+				fputc('<',out); fputc(ch,out); 
+				state=INTXT;
+			} else { 
+				if ( is_rtag(buf,lenbuf) ) {
+					state = INDROPTAG;
+					closelen=0;
+				} else {
+					state=INTAG;
+					fputc(' ',out);
+				}
+			}
+		} else if ( state==INTAG ) {
+			if ( ch == '>' ) {
+				state=INTXT;
+				fputc(' ',out);
+			}	
+		} else if ( state == INDROPTAG ) {
+			if ( ch == '<' ) {
+				state=INDROPINTAG;
+				closelen=0;
+			}
+		} else if ( state == INDROPINTAG ) {
+			if ( ch == '/' )
+				state=INDROPINCLSTAG;
+			else
+				state=INDROPTAG;
+		} else if ( state == INDROPINCLSTAG ) {
+			if ( isalpha(ch) ) {
+				if ( closelen < lenbuf && tolower(ch) == buf[closelen] ) {
+					closelen++;
+					if ( closelen==lenbuf )
+						state=INTAG;
+				} else 
+					state=INDROPTAG;
+			} else
+				state=INDROPTAG;
+		} else if ( state==FINDEND ) {
+			if ( ch == ' ' ) {
+				buf[ lenbuf ] = ch;
+				lenbuf++;
+				if ( lenbuf > spacelen ) {
+					fwrite(buf, sizeof(char), lenbuf, out);
+					state=INTXT;
+				}
+			} else if ( ch=='\n' ) {
+				buf[ lenbuf ] = ch;
+				lenbuf++;
+				fwrite(buf, sizeof(char), lenbuf, out);
+				state=INTXT;
+			} else if ( ch !='\r' ) {
+				state=INTXT;
+				fputc(' ',out);
+				ungetc(ch,in);
+			} 
+		} else {
+			printf("Unknown state: %d\n", state);
+			exit(1);
+		}
+	} 
+			
+	if ( in!=stdin )
+		fclose(in);
+	if ( out!=stdout )
+		fclose(out);
+
+	return 0;	
+}
+