hhc.c

又一开源的解压缩chm格式文件的代码

/***************************************************************************
 *   Copyright (C) 2003 by zhong                                           *
 *   [e-mail]                                                        *
 *                                                                         *
 *   This program is free software; you can redistribute it and/or modify  *
 *   it under the terms of the GNU General Public License as published by  *
 *   the Free Software Foundation; either version 2 of the License, or     *
 *   (at your option) any later version.                                   *
 ***************************************************************************/

#include <sys/types.h>
//#include <sys/stat.h>
#include <unistd.h>
#include <string.h>
#include <stdio.h>
#include <stdlib.h>
//#include <dirent.h>
#include <gtk/gtk.h>
#include "interface.h"
#include "hhc.h"

static gint getTagNumber( const gchar *data );
static gchar *getTag( const gchar *data, int index );
static gchar *getNameFromTag( const gchar *data );
static gchar *getParamValueFromTag( const gchar *data, const gchar *paramName );
static int LoadHhc (ChmSee *window, FILE *hhc);
// static int IsDirectory(char *buffer);

/*
 *
 *  下面这些代码来自 chm_viewer 程序，但我作了修改，让本程序使用。
 *  感谢 chm_viewer 作者 zbw(朱等翔) 的工作。
 *
 */
static gint 
getTagNumber( const gchar *data ) {
	gint length = strlen( data );

    gint label = 0;
    gint num = 0;
	gint i;

    for( i=0; i<length; i++ ) {
		if( data[i]=='<' ) {
			label = 1;
        } else if( data[i]=='>' ) {
			if( label==1 ) {
            	num++;
            	label = 0;
            }
        }
    }

    return num;
}

static gchar *
getTag( const gchar *data, int index ) {
	int length = strlen( data );
    char buffer[512];
	gchar *sendbuf;

	if( index<=0 || index > getTagNumber( data ) ) {
		
		sendbuf = g_strdup("\"\"");
    	return ( sendbuf );
	}

	int label = 0;
    int num = 0;
    int loop =0;
	int i;

    for( i=0; i<length; i++ ) {
		switch( data[i] ) {
			case '<':
            	label = 1;
                num++;

            	break;
            case '>':
            	label = 0;

            	break;
        }

        if( label==1 && num==index ) {
            buffer[loop] = data[i];
			loop++;
        }

        if( label==0 && num==index ) {
			buffer[loop] = '>';
            buffer[loop + 1] = '\0';

            break;
        }
    }
	
	sendbuf = (gchar *) buffer;
    return sendbuf;
}

static gchar *
getNameFromTag( const gchar *data ) {
	char buffer[512];
	
	gchar *sendbuf;

    /**
     * 控制当前位置在 tag 中的位置的表示
     * 0:初始状态
     * 1:进入<并且在tag内的第一个字符前
     * 2:在  tag name 内
     * 3:离开 tag name
     *
     * < /  tag >:5表示在 / 上
     *          4表示在 / 和 tag 之间
     * / 和 tag间的空格需要过滤掉，用 /tag表示这是一个 tag 标签的关闭标签
     */
    int label = 0;
    int loop = 0;

	int length = strlen( data );
	int i;
    for( i=0; i<length; i++ ) {
		switch( data[i] ) {
			case '<':
            	if( label==0 )
            		label = 1;
            	break;
            case ' ':
            	if( label==2 )
                	label = 3;
            	if( label==5 )
                	label = 4;

                break;
            case '/':
            	if( label==2 )
                	label = 3;
                if( label==1 )
					label = 5;

            	break;
            case '>':
            	if( label==2 )
                	label = 3;
            	break;
            default:
            	if( label==1 || label==4 || label==5 )
					label = 2;

            	break;
        }

        if( label==3 ) {
			buffer[loop] = '\0';
        	break;
        }

        //是 name 的字符，添加
        if( label==2 || label==5 ) {
        	buffer[loop] = data[i];
            loop ++;
        }
    }

    sendbuf = buffer;
    return sendbuf;
}


static gchar *
getParamValueFromTag( const gchar *data, const gchar *paramName ) {
    
    char paramNameBuffer[255];
    char paramValueBuffer[512];

    int nameLoop = 0;
    int valueLoop = 0;
    gchar *sendbuffer;

    /**
     * 控制当前位置在 tag 中的位置的表示
     * 0:初始状态
     * 1:进入<并且在tag内的第一个字符前
     * 2:在  tag name 内
     * 3:结束tag
     *
     * < /  tag >:5表示在 / 上
     *          4表示在 / 和 tag 之间
     * / 和 tag间的空格需要过滤掉，用 /tag表示这是一个 tag 标签的关闭标签
     * 11:离开 name 进入参数区
     * 12:进入参数名称
     * 13:离开参数名称
     * 14:进入参数名称后面的等号
     * 16:在＝后面的"上
     * 15:进入参数的value区域
     * value提取完后重新进入 11 状态
     */
    int label = 0;

    /**
     * 0:表示参数的value没有使用""引起来
     * 1:表示参数的value使用""引起来了
     */
    int valueType = 0;

    int length = strlen( data );
    int i;
 
    for( i=0; i<length; i++ ) {
		switch( data[i] ) {
			case '<':
            	if( label==0 )
            		label = 1;

                if( label==16 ) {
                	label = 15;
                }
            	break;
            case ' ':
            	if( label==2 )
                	label = 11;
            	if( label==5 )
                	label = 4;

            	if( label==12 )
                	label = 13;

                if( label==16 ) {
                	label = 15;
                }

                if( label==15 && valueType==0) {
                	//离开前一个参数的时候，需要把参数名称清空
					paramNameBuffer[0] = '\0';
                    nameLoop = 0;

                	label = 11;
                }
                break;
            case '/':
            	if( label==2 )
                	label = 3;
                if( label==1 )
					label = 5;

                if( label==16 ) {
                	label = 15;
                }

            	break;
            case '>':
                if( label==16 ) {
                	label = 15;
                }

            	if( valueType==0 )
                	label = 3;
                else {
					if( label!=15 )
                    	label = 3;
                }

            	break;
            case '=':
            	if( label==12 || label==13 )
                	label = 14;

                if( label==16 ) {
                	label = 15;
                }
            	break;
            case '\"':
                if( label==15 || label==16 ) {
                	//离开前一个参数的时候，需要把参数名称清空
					paramNameBuffer[0] = '\0';
                    nameLoop = 0;
                	label = 11;
                }

                if( label==14 ) {
                	label = 16;
					valueType = 1;
                }

            	break;
            default:
            	if( label==1 || label==4 || label==5 )
					label = 2;

                if( label==11 ) {
					//在参数前面的空格区域
                    label = 12;
                }

                if( label==14 ) {
                	label = 15;
					valueType = 0;
                }

                if( label==16 ) {
                	label = 15;
                }

            	break;
        }

        if( label==5 || label==3) {
			//结束标签，不能有参数
            break;
        }

        if( label==12 ) {
			//提取参数名称
			paramNameBuffer[nameLoop] = data[i];
            paramNameBuffer[nameLoop + 1] = '\0';
            nameLoop++;
        }

	//printf( "label:%d========char:%c========param:%s\n", label, data[i], paramNameBuffer );

        if( label==15 ) {
			//在value区域内，判断是否是需要的value
            if( g_strcasecmp(paramName, g_strdown(paramNameBuffer)) == 0) {
		paramValueBuffer[valueLoop] = data[i];
                paramValueBuffer[valueLoop + 1] = '\0';
		valueLoop++;
            }
        }
    }

	//printf( "loop:%d============value:%s=========data:%s\n", valueLoop, paramValueBuffer, data );

    if( valueLoop==0 )
    	paramValueBuffer[valueLoop] = '\0';

    sendbuffer = paramValueBuffer;
    return sendbuffer;
}

static int
LoadHhc (ChmSee *window, FILE *hhc)
{
	char	lineBuf[256];
	GtkTreeView		*tree = GTK_TREE_VIEW(window->chm_tree);
	GtkTreeModel	*model = gtk_tree_view_get_model (tree);
	GtkTreeIter iter;
	GtkTreeIter	sibling_iter;
	gchar	*tagData;
	gchar	*tmpName;
	gchar	*tmpValue;
	gchar	*currentName = NULL;
	gchar	*local = NULL;
	gchar	*title_utf8; // = NULL;
	gchar	*title; // = NULL;
	gchar	*local_conv; // = NULL;
	gint	label = 0;
	gint	doc_label = 0;
	gint	last_label = 0;
	gint	tagNum;
	gint	j;

	while (fgets (lineBuf, 512, hhc) != NULL)
	{
		tagNum = getTagNumber( lineBuf );
        
		for( j=1; j<=tagNum; j++ ) {

        	tagData = g_strdup(getTag( lineBuf, j ));
			tmpName = g_strdup(g_strdown(getNameFromTag( tagData )));

			if( g_strcasecmp("ul", tmpName) == 0) {
				//进入一个 UL
				label = 1;
				doc_label++;

			} else if( g_strcasecmp("li", tmpName) == 0) {
            	local = NULL;
               	currentName = NULL;
				title_utf8 = NULL;
				title = NULL;
				local_conv = NULL;
				label = 3;

        	} else if( g_strcasecmp("param", tmpName) == 0) {
				if( label==3 ) {
		//记录参数
            		tmpValue = g_strdup(getParamValueFromTag(tagData, "name"));
            		if( g_strcasecmp("name", g_strdown(tmpValue)) == 0 ) {
						currentName = g_strdup( getParamValueFromTag( tagData, "value" ));
                	} else if( g_strcasecmp("local", g_strdown(tmpValue)) == 0 ) {
						local = g_strdup( getParamValueFromTag( tagData, "value" ));
                	}
             	}
         	} else if( g_strcasecmp("/object", tmpName) == 0) {
            	if( label==3 ) {
					//生成目录条目
					if (local == NULL) {
						local = g_strdup_printf("%s", "nolink");
					}
					
					title_utf8 = g_convert(currentName, -1, "UTF-8", "GB18030", NULL, NULL, NULL);
					
					if (title_utf8) {
						title = g_strdup (title_utf8);
					} else {
						title = g_strdup(currentName);
					}
					
					local_conv = g_strdup(local);
					
					if (doc_label == 1) {

						gtk_tree_store_append(GTK_TREE_STORE (model), &iter, NULL);
						gtk_tree_store_set (GTK_TREE_STORE (model),
				    						&iter,
				    						TITLE_COLUMN, title,
				    						FILENAME_COLUMN, local_conv,
				    						-1);
						sibling_iter = iter;
						last_label = 1;
						
					} else  {
					
						if ((doc_label - last_label) == 1) {

							gtk_tree_store_append(GTK_TREE_STORE (model), &iter, &sibling_iter);
							gtk_tree_store_set (GTK_TREE_STORE (model),
				    						&iter,
				    						TITLE_COLUMN, title,
				    						FILENAME_COLUMN, local_conv,
				    						-1);
							sibling_iter = iter;
							last_label++;

						} else if ((doc_label - last_label) == 0) {
							
							gtk_tree_store_insert_after(GTK_TREE_STORE (model), &iter, NULL, &sibling_iter);
							gtk_tree_store_set (GTK_TREE_STORE (model),
				    						&iter,
				    						TITLE_COLUMN, title,
				    						FILENAME_COLUMN, local_conv,
				    						-1);
							sibling_iter = iter;
							last_label = doc_label;

						} else if  ((doc_label - last_label)== -1) {

							gtk_tree_store_insert_after(GTK_TREE_STORE (model), &iter, NULL, &sibling_iter);
							gtk_tree_store_set (GTK_TREE_STORE (model),
				    						&iter,
				    						TITLE_COLUMN, title,
				    						FILENAME_COLUMN, local_conv,
				    						-1);
							sibling_iter = iter;
							last_label--;							

						} else {
							
							g_print("Error\n");
						}
					}
					
                }

			} else if( g_strcasecmp("/ul", tmpName) == 0) {
				//向上返回一个级别
				gtk_tree_model_iter_parent(model, &iter, &sibling_iter);
				sibling_iter = iter;
			 	doc_label--;
				last_label--;
			}
		}
	}

//	g_free(currentName);
//	g_free(local);
//	g_free(title_utf8);
//	g_free(title);
//	g_free(local_conv);

	return TRUE;
}

/*
 * 到这里结束
 */
//int open_hhc_file (ChmSee *window, const gchar *filename)
int
open_hhc_file (ChmSee *window)
{
	FILE	*hhc;

	hhc = fopen (window->hhc_name, "r");

	if (!hhc) {
		return FALSE;
	}
	
	if(!LoadHhc(window, hhc)) {
		return FALSE;
	}

	fclose (hhc);
	return TRUE;
}