* use OpenCV rather than Imlib2 if both are available.

* cleaned up the slashdot code.
  * decode phpBB captchas.
  * added filter_contrast.


git-svn-id: file:///srv/caca.zoy.org/var/lib/svn/pwntcha/trunk@389 92316355-f0b4-4df1-b90c-862c8a59935f

configure.ac

@@ -18,6 +18,19 @@ AC_C_CONST
 AC_C_INLINE
 AC_TYPE_SIZE_T
 
+# Use OpenCV?
+ac_cv_my_have_opencv="no"
+save_CPPFLAGS="${CPPFLAGS}"
+AC_PATH_PROG(OPENCV_CONFIG, opencv-config, no)
+if test "${OPENCV_CONFIG}" != "no"; then
+  CPPFLAGS="${CPPFLAGS} `opencv-config --cflags`"
+fi
+AC_CHECK_HEADERS(cv.h,
+ [ac_cv_my_have_opencv="yes"],
+ [ac_cv_my_have_opencv="no"])
+CPPFLAGS="${save_CPPFLAGS}"
+AM_CONDITIONAL(USE_OPENCV, test "${ac_cv_my_have_opencv}" = "yes")
+
 # Use Imlib2?
 ac_cv_my_have_imlib2="no"
 save_CPPFLAGS="${CPPFLAGS}"
@@ -31,19 +44,6 @@ AC_CHECK_HEADERS(Imlib2.h,
 CPPFLAGS="${save_CPPFLAGS}"
 AM_CONDITIONAL(USE_IMLIB2, test "${ac_cv_my_have_imlib2}" = "yes")
 
-# Use Imlib2?
-ac_cv_my_have_opencv="no"
-save_CPPFLAGS="${CPPFLAGS}"
-AC_PATH_PROG(OPENCV_CONFIG, opencv-config, no)
-if test "${OPENCV_CONFIG}" != "no"; then
-  CPPFLAGS="${CPPFLAGS} `opencv-config --cflags`"
-fi
-AC_CHECK_HEADERS(cv.h,
- [ac_cv_my_have_opencv="yes"],
- [ac_cv_my_have_opencv="no"])
-CPPFLAGS="${save_CPPFLAGS}"
-AM_CONDITIONAL(USE_OPENCV, test "${ac_cv_my_have_opencv}" = "yes")
-
 if test "${ac_cv_my_have_imlib2}" = "no" -a "${ac_cv_my_have_opencv}" = "no"; then
   AC_MSG_ERROR([[cannot find Imlib2 or OpenCV, please install one of them]])
 fi

src/Makefile.am

@@ -3,16 +3,16 @@ NULL =
 bin_PROGRAMS = pwntcha
 pwntcha_CFLAGS = $(ADDITIONAL_CFLAGS) -Wall -O6
 pwntcha_LDFLAGS = $(ADDITIONAL_LDFLAGS)
-pwntcha_SOURCES = main.c image.c filters.c slashdot.c common.h
+pwntcha_SOURCES = main.c image.c filters.c common.h slashdot.c phpbb.c test.c
 
-if USE_IMLIB2
-ADDITIONAL_CFLAGS = `imlib2-config --cflags` -DX_DISPLAY_MISSING=1
-ADDITIONAL_LDFLAGS = `imlib2-config --libs`
-else
 if USE_OPENCV
 ADDITIONAL_CFLAGS = `opencv-config --cflags`
 ADDITIONAL_LDFLAGS = `opencv-config --libs opencv highgui`
 else
+if USE_IMLIB2
+ADDITIONAL_CFLAGS = `imlib2-config --cflags` -DX_DISPLAY_MISSING=1
+ADDITIONAL_LDFLAGS = `imlib2-config --libs`
+else
 ADDITIONAL_CFLAGS =
 ADDITIONAL_LDFLAGS =
 endif

src/common.h

@@ -18,11 +18,15 @@ struct image
 };
 
 /* available CAPTCHA decoders */
-char * decode_slashdot(struct image *img);
+char *decode_phpbb(struct image *img);
+char *decode_slashdot(struct image *img);
+char *decode_test(struct image *img);
 
 /* image operations */
-struct image * image_load(char *name);
-struct image * image_new(int width, int height);
+struct image *image_load(char *name);
+struct image *image_new(int width, int height);
+void image_free(struct image *img);
+void image_display(struct image *img);
 int getgray(struct image *img, int x, int y, int *g);
 int getpixel(struct image *img, int x, int y, int *r, int *g, int *b);
 int setpixel(struct image *img, int x, int y, int r, int g, int b);
@@ -31,8 +35,9 @@ int setpixel(struct image *img, int x, int y, int r, int g, int b);
 void filter_flood_fill(struct image *img, int x, int y, int r, int g, int b);
 struct image *filter_fill_holes(struct image *img);
 struct image *filter_detect_lines(struct image *img);
-struct image *filter_equalize(struct image *img);
+struct image *filter_equalize(struct image *img, int threshold);
 struct image *filter_trick(struct image *img);
 struct image *filter_smooth(struct image *img);
 struct image *filter_median(struct image *img);
+struct image *filter_contrast(struct image *img);
 

src/filters.c

@@ -142,7 +142,7 @@ struct image *filter_detect_lines(struct image *img)
     return dst;
 }
 
-struct image *filter_equalize(struct image *img)
+struct image *filter_equalize(struct image *img, int threshold)
 {
     struct image *dst;
     int x, y;
@@ -154,7 +154,7 @@ struct image *filter_equalize(struct image *img)
         for(x = 0; x < img->width; x++)
         {
             getpixel(img, x, y, &r, &g, &b);
-            if(r < 200) r = 50; else r = 200;
+            if(r < threshold) r = 0; else r = 255;
             setpixel(dst, x, y, r, r, r);
         }
 
@@ -240,7 +240,7 @@ struct image *filter_smooth(struct image *img)
 
 struct image *filter_median(struct image *img)
 {
-#define MSIZE 4
+#define MSIZE 3
     struct image *dst;
     int x, y, i, j, val[MSIZE*MSIZE];
     int r, g, b;
@@ -278,3 +278,36 @@ struct image *filter_median(struct image *img)
     return dst;
 }
 
+struct image *filter_contrast(struct image *img)
+{
+    struct image *dst;
+    int histo[256];
+    int x, y, i, min = 255, max = 0;
+    int r, g, b;
+
+    dst = image_new(img->width, img->height);
+
+    for(y = 0; y < img->height; y++)
+        for(x = 0; x < img->width; x++)
+        {
+            getgray(img, x, y, &r);
+            if(r < min) min = r;
+            if(r > max) max = r;
+        }
+
+    if(min == max)
+        histo[min] = 127;
+    else
+        for(i = min; i < max; i++)
+            histo[i] = (i - min) * 255 / (max - min);
+
+    for(y = 0; y < img->height; y++)
+        for(x = 0; x < img->width; x++)
+        {
+            getgray(img, x, y, &r);
+            setpixel(dst, x, y, histo[r], histo[r], histo[r]);
+        }
+
+    return dst;
+}
+

src/image.c

@@ -15,79 +15,93 @@
 #include "config.h"
 #include "common.h"
 
-#if defined(HAVE_IMLIB2_H)
-#   include <Imlib2.h>
-#elif defined(HAVE_CV_H)
+#if defined(HAVE_CV_H)
 #   include <cv.h>
 #   include <highgui.h>
+#elif defined(HAVE_IMLIB2_H)
+#   include <Imlib2.h>
 #else
 #   error "No imaging library"
 #endif
 
-struct image * image_load(char *name)
+struct image *image_load(char *name)
 {
-    struct image * img;
-#if defined(HAVE_IMLIB2_H)
+    struct image *img;
+#if defined(HAVE_CV_H)
+    IplImage *priv = cvLoadImage(name, -1);
+#elif defined(HAVE_IMLIB2_H)
     Imlib_Image priv = imlib_load_image(name);
-#elif defined(HAVE_CV_H)
-    IplImage * priv = cvLoadImage(name, -1);
 #endif
 
     if(!priv)
         return NULL;
 
     img = malloc(sizeof(struct image));
-#if defined(HAVE_IMLIB2_H)
+#if defined(HAVE_CV_H)
+    img->width = priv->width;
+    img->height = priv->height;
+    img->pitch = priv->widthStep;
+    img->channels = priv->nChannels;
+    img->pixels = priv->imageData;
+#elif defined(HAVE_IMLIB2_H)
     imlib_context_set_image(priv);
     img->width = imlib_image_get_width();
     img->height = imlib_image_get_height();
     img->pitch = 4 * imlib_image_get_width();
     img->channels = 4;
     img->pixels = (char *)imlib_image_get_data();
-#elif defined(HAVE_CV_H)
-    img->width = priv->width;
-    img->height = priv->height;
-    img->pitch = priv->widthStep;
-    img->channels = priv->nChannels;
-    img->pixels = priv->imageData;
 #endif
     img->priv = (void *)priv;
 
     return img;
 }
 
-struct image * image_new(int width, int height)
+struct image *image_new(int width, int height)
 {
-    struct image * img;
-#if defined(HAVE_IMLIB2_H)
+    struct image *img;
+#if defined(HAVE_CV_H)
+    IplImage *priv = cvCreateImage(cvSize(width, height), IPL_DEPTH_8U, 3);
+#elif defined(HAVE_IMLIB2_H)
     Imlib_Image priv = imlib_create_image(width, height);
-#elif defined(HAVE_CV_H)
-    IplImage * priv = cvCreateImage(cvSize(width, height), IPL_DEPTH_8U, 3);
 #endif
 
     if(!priv)
         return NULL;
 
     img = malloc(sizeof(struct image));
-#if defined(HAVE_IMLIB2_H)
+#if defined(HAVE_CV_H)
+    img->width = priv->width;
+    img->height = priv->height;
+    img->pitch = priv->widthStep;
+    img->channels = priv->nChannels;
+    img->pixels = priv->imageData;
+#elif defined(HAVE_IMLIB2_H)
     imlib_context_set_image(priv);
     img->width = imlib_image_get_width();
     img->height = imlib_image_get_height();
     img->pitch = 4 * imlib_image_get_width();
     img->channels = 4;
     img->pixels = (char *)imlib_image_get_data();
-#elif defined(HAVE_CV_H)
-    img->width = priv->width;
-    img->height = priv->height;
-    img->pitch = priv->widthStep;
-    img->channels = priv->nChannels;
-    img->pixels = priv->imageData;
 #endif
     img->priv = (void *)priv;
 
     return img;
 }
 
+void image_free(struct image *img)
+{
+#if defined(HAVE_CV_H)
+    IplImage *iplimg;
+    iplimg = (IplImage *)img->priv;
+    cvReleaseImage(&iplimg);
+#elif defined(HAVE_IMLIB2_H)
+    imlib_context_set_image(img->priv);
+    imlib_free_image();
+#endif
+
+    free(img);
+}
+
 int getgray(struct image *img, int x, int y, int *g)
 {
     if(x < 0 || y < 0 || x >= img->width || y >= img->height)
@@ -132,19 +146,21 @@ int setpixel(struct image *img, int x, int y, int r, int g, int b)
 
 void image_display(struct image *img)
 {
-#if defined(HAVE_IMLIB2_H)
+#if defined(HAVE_CV_H)
+    char name[BUFSIZ];
+    sprintf(name, "Image %p (%i x %i)", img, img->width, img->height);
+    cvNamedWindow(name, 0);
+    cvShowImage(name, img->priv);
+    cvResizeWindow(name, img->width * 2, img->height * 2 + 50);
+    while((unsigned char)cvWaitKey(0) != 0x1b)
+        ;
+#elif defined(HAVE_IMLIB2_H)
     //char name[BUFSIZ];
     //static int i = 0;
     //sprintf(name, "image%i-%ix%i.png", i++, img->width, img->height);
     //imlib_context_set_image(img->priv);
     //imlib_save_image(name);
     //fprintf(stderr, "saved to %s\n", name);
-#elif defined(HAVE_CV_H)
-    char name[BUFSIZ];
-    sprintf(name, "Image %p (%i x %i)", img, img->width, img->height);
-    cvNamedWindow(name, 0);
-    cvShowImage(name, img->priv);
-    cvResizeWindow(name, 320, 120);
 #endif
 }
 

src/main.c

@@ -11,6 +11,7 @@
 
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
 #include <getopt.h>
 
 #include "config.h"
@@ -87,7 +88,29 @@ int main(int argc, char *argv[])
             continue;
         }
 
-        result = decode_slashdot(img);
+        if(!strcmp(mode, "test"))
+            result = decode_test(img);
+        else if(!strcmp(mode, "phpbb"))
+            result = decode_phpbb(img);
+        else if(!strcmp(mode, "slashdot"))
+            result = decode_slashdot(img);
+        else
+        {
+            if(img->width == 320 && img->height == 50)
+                result = decode_phpbb(img);
+            else if(img->height == 69)
+                result = decode_slashdot(img);
+            else
+            {
+                fprintf(stderr, "%s: could not guess CAPTCHA type\n", argv[0]);
+                printf("\n");
+                image_free(img);
+                continue;
+            }
+        }
+
+        image_free(img);
+
         if(!result)
         {
             fprintf(stderr, "%s: sorry, decoding failed\n", argv[0]);

src/phpbb.c

@@ -0,0 +1,136 @@
+/*
+ * phpbb.c: decode phpBB captchas
+ * $Id$
+ *
+ * Copyright: (c) 2005 Sam Hocevar <sam@zoy.org>
+ *   This program is free software; you can redistribute it and/or
+ *   modify it under the terms of the Do What The Fuck You Want To
+ *   Public License as published by Banlu Kemiyatorn. See
+ *   http://sam.zoy.org/projects/COPYING.WTFPL for more details.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "config.h"
+#include "common.h"
+
+/* Our macros */
+#define FONTNAME "share/font_phpbb.png"
+
+static struct image *find_glyphs(struct image *img);
+
+/* Global stuff */
+char *result;
+
+/* Main function */
+char *decode_phpbb(struct image *img)
+{
+    char all[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ123456789";
+    struct image *tmp1, *tmp2, *tmp3;
+    struct image *font = image_load(FONTNAME);
+    int x, y, i = 0;
+    int r, g, b;
+    int xmin, xmax, ymin, ymax, cur, offset = -1;
+    int distmin, distx, disty, distch;
+
+    /* phpBB captchas have 6 characters */
+    result = malloc(7 * sizeof(char));
+
+    if(!font)
+    {
+        fprintf(stderr, "cannot load font %s\n", FONTNAME);
+        exit(-1);
+    }
+
+    tmp1 = filter_smooth(img);
+    tmp2 = filter_equalize(tmp1, 128);
+    tmp3 = image_new(img->width, img->height);
+
+    for(x = 0; x < img->width; x++)
+        for(y = 0; y < img->height; y++)
+        {
+            getpixel(tmp2, x, y, &r, &g, &b);
+            if(r == 0 && offset == -1)
+                offset = x;
+            getpixel(img, x, y, &r, &g, &b);
+            setpixel(tmp3, x, y, 255, g, 255);
+        }
+
+    strcpy(result, "       ");
+
+    for(cur = 0; cur < 6; cur++)
+    {
+        /* Try to find 1st letter */
+        distmin = 999999999;
+        for(i = 0; i < 35; i++)
+        {
+            int localmin = 99999999, localx, localy;
+            xmin = i * 40;
+            ymin = 0;
+            xmax = i * 40 + 40;
+            ymax = 40;
+            for(y = 0; y < img->height - 40; y++)
+            {
+                x = offset - 3;
+                if(cur == 0)
+                    x -= 10;
+                if(x < 0)
+                    x = 0;
+                for(; x < offset + 3; x++)
+                {
+                    int z, t, dist;
+                    dist = 0;
+                    for(t = 0; t < ymax - ymin; t++)
+                        for(z = 0; z < xmax - xmin; z++)
+                        {
+                            int r2;
+                            getgray(font, xmin + z, ymin + t, &r);
+                            getgray(tmp2, x + z, y + t, &r2);
+                            dist += (r - r2) * (r - r2);
+                        }
+                    if(dist < localmin)
+                    {
+                        localmin = dist;
+                        localx = x;
+                        localy = y;
+                    }
+                }
+            }
+            if(localmin < distmin)
+            {
+                distmin = localmin;
+                distx = localx;
+                disty = localy;
+                distch = i;
+            }
+        }
+
+        /* Print min glyph (debug) */
+        xmin = distch * 40;
+        ymin = 0;
+        xmax = distch * 40 + 40;
+        ymax = 40;
+        for(y = 0; y < ymax - ymin; y++)
+            for(x = 0; x < xmax - xmin; x++)
+            {
+                int r2;
+                getpixel(font, xmin + x, ymin + y, &r2, &g, &b);
+                if(r2 > 128) continue;
+                getpixel(tmp3, distx + x, disty + y, &r, &g, &b);
+                setpixel(tmp3, distx + x, disty + y, r2, g, b);
+            }
+
+        offset = distx + xmax - xmin;
+        result[cur] = all[distch];
+    }
+
+    image_free(tmp1);
+    image_free(tmp2);
+    image_free(tmp3);
+    image_free(font);
+
+    return result;
+}
+

src/slashdot.c

@@ -30,30 +30,48 @@ static struct image *find_glyphs(struct image *img);
 
 /* Global stuff */
 struct { int xmin, ymin, xmax, ymax; } objlist[100];
-int objects = 0, first = -1, last = -1;
+int objects, first, last;
 char *result;
 
 /* Main function */
 char * decode_slashdot(struct image *img)
 {
-    struct image *tmp, *tmp2;
+    struct image *tmp1, *tmp2, *tmp3, *tmp4, *tmp5, *tmp6, *tmp7;
+
+    /* Initialise local data */
+    objects = 0;
+    first = -1;
+    last = -1;
 
     /* Slashdot captchas have 7 characters */
     result = malloc(8 * sizeof(char));
 
     /* Clean image a bit */
-    tmp = filter_detect_lines(img);
-    tmp = filter_fill_holes(tmp);
+    tmp1 = filter_detect_lines(img);
+    tmp2 = filter_fill_holes(tmp1);
 
     /* Detect small objects to guess image orientation */
-    tmp2 = filter_median(tmp);
-    tmp2 = filter_equalize(tmp2);
-    count_objects(tmp2);
+    tmp3 = filter_median(tmp2);
+    tmp4 = filter_equalize(tmp3, 200);
+    count_objects(tmp4);
 
     /* Invert rotation and find glyphs */
-    tmp = rotate(tmp);
-    tmp = filter_median(tmp);
-    tmp = find_glyphs(tmp);
+    tmp5 = rotate(tmp2);
+    tmp6 = filter_median(tmp5);
+    tmp7 = find_glyphs(tmp6);
+
+    /* Clean up our mess */
+    image_free(tmp1);
+    image_free(tmp2);
+    image_free(tmp3);
+    image_free(tmp4);
+    image_free(tmp5);
+    image_free(tmp6);
+    image_free(tmp7);
+
+    /* aaaaaaa means decoding failed */
+    if(!strcmp(result, "aaaaaaa"))
+        result[0] = '\0';
 
     return result;
 }

src/test.c

@@ -0,0 +1,155 @@
+/*
+ * test.c: test captchas
+ * $Id$
+ *
+ * Copyright: (c) 2004 Sam Hocevar <sam@zoy.org>
+ *   This program is free software; you can redistribute it and/or
+ *   modify it under the terms of the Do What The Fuck You Want To
+ *   Public License as published by Banlu Kemiyatorn. See
+ *   http://sam.zoy.org/projects/COPYING.WTFPL for more details.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <math.h>
+
+#include "config.h"
+#include "common.h"
+
+/* Our macros */
+#define FONTNAME "share/font_phpbb.png"
+
+static struct image *find_glyphs(struct image *img);
+
+/* Global stuff */
+struct { int xmin, ymin, xmax, ymax; } objlist[100];
+int objects, first, last;
+char *result;
+
+/* Main function */
+char * decode_test(struct image *img)
+{
+    struct image *tmp1, *tmp2, *tmp3, *tmp4, *tmp5, *tmp6, *tmp7;
+
+    /* Initialise local data */
+    objects = 0;
+    first = -1;
+    last = -1;
+
+    /* phpBB captchas have 6 characters */
+    result = malloc(7 * sizeof(char));
+
+    tmp1 = filter_smooth(img);
+    tmp2 = filter_median(tmp1);
+    tmp3 = filter_equalize(tmp2, 130);
+    tmp4 = filter_median(tmp3);
+    tmp5 = find_glyphs(tmp3);
+
+    image_free(tmp1);
+    image_free(tmp2);
+    image_free(tmp3);
+    image_free(tmp4);
+    image_free(tmp5);
+
+    return result;
+}
+
+/* The following functions are local */
+
+static struct image *find_glyphs(struct image *img)
+{
+    char all[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ123456789";
+    struct image *dst;
+    struct image *font = image_load(FONTNAME);
+    int x, y, i = 0;
+    int r, g, b;
+    int xmin, xmax, ymin, ymax, incell = 0, count = 0, cur = 0, offset = -1;
+    int distmin, distx, disty, distch;
+
+    if(!font)
+    {
+        fprintf(stderr, "cannot load font %s\n", FONTNAME);
+        exit(-1);
+    }
+
+    dst = image_new(img->width, img->height);
+
+    for(x = 0; x < img->width; x++)
+        for(y = 0; y < img->height; y++)
+        {
+            getpixel(img, x, y, &r, &g, &b);
+            setpixel(dst, x, y, 255, g, 255);
+            if(r == 0 && offset == -1)
+                offset = x;
+        }
+
+    strcpy(result, "       ");
+
+    while(cur < 6)
+    {
+        /* Try to find 1st letter */
+        distmin = 999999999;
+        for(i = 0; i < 35; i++)
+        {
+            int localmin = 99999999, localx, localy;
+            xmin = i * 40;
+            ymin = 0;
+            xmax = i * 40 + 40;
+            ymax = 40;
+            for(y = 0; y < img->height - 40; y++)
+            {
+                x = offset - 5;
+                if(cur == 0)
+                    x -= 15;
+                if(x < 0)
+                    x = 0;
+                for(; x < offset + 10; x++)
+                {
+                    int z, t, dist;
+                    dist = 0;
+                    for(t = 0; t < ymax - ymin; t++)
+                        for(z = 0; z < xmax - xmin; z++)
+                        {
+                            int r2;
+                            getgray(font, xmin + z, ymin + t, &r);
+                            getgray(img, x + z, y + t, &r2);
+                            dist += abs(r - r2);
+                        }
+                    if(dist < localmin)
+                    {
+                        localmin = dist;
+                        localx = x;
+                        localy = y;
+                    }
+                }
+            }
+            if(localmin < distmin)
+            {
+                distmin = localmin;
+                distx = localx;
+                disty = localy;
+                distch = i;
+            }
+        }
+
+        /* Print min glyph (debug) */
+        xmin = distch * 40;
+        ymin = 0;
+        xmax = distch * 40 + 40;
+        ymax = 40;
+        for(y = 0; y < ymax - ymin; y++)
+            for(x = 0; x < xmax - xmin; x++)
+            {
+                getpixel(font, xmin + x, ymin + y, &r, &g, &b);
+                if(r > 128) continue;
+                setpixel(dst, distx + x, disty + y, r, g, b);
+            }
+
+        offset = distx + xmax - xmin;
+        result[cur++] = all[distch];
+    }
+
+    return dst;
+}
+