Changeset 9e219d3

Timestamp:

Mar 13, 2016, 12:43:12 PM (8 years ago)

Author:

Karl Ramm <kcr@1ts.org>

Children:

ed05467

Parents:

0f5af62

git-author:

Karl Ramm <kcr@1ts.org> (03/12/16 21:40:02)

git-committer:

Karl Ramm <kcr@1ts.org> (03/13/16 12:43:12)

Message:

rewrite the zephyr scribelike markup parser

Files:

• AUTHORS (modified) (1 diff)
• Makefile.am (modified) (1 diff)
• fmtext.c (modified) (2 diffs)
• owl.h (modified) (1 diff)
• tester.c (modified) (3 diffs)
• ztext.c (added)
• ztext.h (added)

AUTHORS

@@ -32 +32 @@
   Betsy Riley
   Robert Jacobs
+  Google Inc.
 
 BarnOwl is based on code from Owl, which was originally primarily

Makefile.am

@@ -56 +56 @@
 CODELIST_SRCS=message.c mainwin.c popwin.c zephyr.c messagelist.c \
      commands.c global.c text.c fmtext.c editwin.c \
-     util.c logging.c \
+     util.c logging.c ztext.c \
      perlconfig.c keys.c functions.c zwrite.c viewwin.c help.c filter.c \
      regex.c history.c view.c dict.c variable.c filterelement.c pair.c \

fmtext.c

@@ -549 +549 @@
 }
 
+/* flattens a ztext parse tree into fmtext */
+static void owl_fmtext_append_ztext_tree(owl_fmtext *f, ztext_env *tree, char attr, short fg)
+{
+  char *s;
+  int i;
+  bool handled = false;
+  const struct {
+    const char *tag;
+    const char mask;
+  } TAGS[] = {
+    {"", 0},
+    {"@", 0},
+    {"@bold", OWL_FMTEXT_ATTR_BOLD},
+    {"@b", OWL_FMTEXT_ATTR_BOLD},
+    {"@italic", OWL_FMTEXT_ATTR_UNDERLINE},
+    {"@i", OWL_FMTEXT_ATTR_UNDERLINE},
+    {NULL, 0},
+  };
+
+  if (tree->closer == '@' && tree->content == NULL) {
+    owl_fmtext_append_attr(f, "@", attr, fg, OWL_COLOR_DEFAULT);
+    return;
+  }
+
+  for (i = 0; TAGS[i].tag != NULL; i++) {
+    if (!strcasecmp(tree->label, TAGS[i].tag)) {
+      attr |= TAGS[i].mask;
+      handled = true;
+      break;
+    }
+  }
+
+  if (!handled) {
+    /* continue the dubious practice of printing codes we don't understand */
+    s = g_strdup_printf("%s%c", tree->label, tree->opener);
+    owl_fmtext_append_attr(f, s, attr, fg, OWL_COLOR_DEFAULT);
+    g_free(s);
+  }
+
+  for (ztext_node *p = tree->content; p != NULL; p = p->next) {
+    if (p->type == ZTEXT_NODE_STRING) {
+      owl_fmtext_append_attr(f, p->string, attr, fg, OWL_COLOR_DEFAULT);
+    } else if (p->type == ZTEXT_NODE_ENV) {
+      if (!strcasecmp(p->env->label, "@color")) {
+        s = ztext_strip(p->env);
+        fg = owl_util_string_to_color(s);
+        if (fg == OWL_COLOR_INVALID)
+          fg = OWL_COLOR_DEFAULT;
+        g_free(s);
+      } else {
+        owl_fmtext_append_ztext_tree(f, p->env, attr, fg);
+      }
+    }
+  }
+
+  if (!handled) {
+    s = g_strdup_printf("%c", tree->closer);
+    owl_fmtext_append_attr(f, s, attr, fg, OWL_COLOR_DEFAULT);
+    g_free(s);
+  }
+}
 
 /* Append the text 'text' to 'f' and interpret the zephyr style
@@ -555 +616 @@
 void owl_fmtext_append_ztext(owl_fmtext *f, const char *text)
 {
-  int stacksize, curattrs, curcolor;
-  const char *ptr, *txtptr, *tmpptr;
-  char *buff;
-  int attrstack[32], chrstack[32], colorstack[32];
-
-  curattrs=OWL_FMTEXT_ATTR_NONE;
-  curcolor=OWL_COLOR_DEFAULT;
-  stacksize=0;
-  txtptr=text;
-  while (1) {
-    ptr=strpbrk(txtptr, "@{[<()>]}");
-    if (!ptr) {
-      /* add all the rest of the text and exit */
-      owl_fmtext_append_attr(f, txtptr, curattrs, curcolor, OWL_COLOR_DEFAULT);
-      return;
-    } else if (ptr[0]=='@') {
-      /* add the text up to this point then deal with the stack */
-      buff=g_new(char, ptr-txtptr+20);
-      strncpy(buff, txtptr, ptr-txtptr);
-      buff[ptr-txtptr]='\0';
-      owl_fmtext_append_attr(f, buff, curattrs, curcolor, OWL_COLOR_DEFAULT);
-      g_free(buff);
-
-      /* update pointer to point at the @ */
-      txtptr=ptr;
-
-      /* now the stack */
-
-      /* if we've hit our max stack depth, print the @ and move on */
-      if (stacksize==32) {
-        owl_fmtext_append_attr(f, "@", curattrs, curcolor, OWL_COLOR_DEFAULT);
-        txtptr++;
-        continue;
-      }
-
-      /* if it's an @@, print an @ and continue */
-      if (txtptr[1]=='@') {
-        owl_fmtext_append_attr(f, "@", curattrs, curcolor, OWL_COLOR_DEFAULT);
-        txtptr+=2;
-        continue;
-      }
-        
-      /* if there's no opener, print the @ and continue */
-      tmpptr=strpbrk(txtptr, "(<[{ ");
-      if (!tmpptr || tmpptr[0]==' ') {
-        owl_fmtext_append_attr(f, "@", curattrs, curcolor, OWL_COLOR_DEFAULT);
-        txtptr++;
-        continue;
-      }
-
-      /* check what command we've got, push it on the stack, start
-         using it, and continue ... unless it's a color command */
-      buff=g_new(char, tmpptr-ptr+20);
-      strncpy(buff, ptr, tmpptr-ptr);
-      buff[tmpptr-ptr]='\0';
-      if (!strcasecmp(buff, "@bold")) {
-        attrstack[stacksize]=OWL_FMTEXT_ATTR_BOLD;
-        chrstack[stacksize]=tmpptr[0];
-        colorstack[stacksize]=curcolor;
-        stacksize++;
-        curattrs|=OWL_FMTEXT_ATTR_BOLD;
-        txtptr+=6;
-        g_free(buff);
-        continue;
-      } else if (!strcasecmp(buff, "@b")) {
-        attrstack[stacksize]=OWL_FMTEXT_ATTR_BOLD;
-        chrstack[stacksize]=tmpptr[0];
-        colorstack[stacksize]=curcolor;
-        stacksize++;
-        curattrs|=OWL_FMTEXT_ATTR_BOLD;
-        txtptr+=3;
-        g_free(buff);
-        continue;
-      } else if (!strcasecmp(buff, "@i")) {
-        attrstack[stacksize]=OWL_FMTEXT_ATTR_UNDERLINE;
-        chrstack[stacksize]=tmpptr[0];
-        colorstack[stacksize]=curcolor;
-        stacksize++;
-        curattrs|=OWL_FMTEXT_ATTR_UNDERLINE;
-        txtptr+=3;
-        g_free(buff);
-        continue;
-      } else if (!strcasecmp(buff, "@italic")) {
-        attrstack[stacksize]=OWL_FMTEXT_ATTR_UNDERLINE;
-        chrstack[stacksize]=tmpptr[0];
-        colorstack[stacksize]=curcolor;
-        stacksize++;
-        curattrs|=OWL_FMTEXT_ATTR_UNDERLINE;
-        txtptr+=8;
-        g_free(buff);
-        continue;
-      } else if (!strcasecmp(buff, "@")) {
-        attrstack[stacksize]=OWL_FMTEXT_ATTR_NONE;
-        chrstack[stacksize]=tmpptr[0];
-        colorstack[stacksize]=curcolor;
-        stacksize++;
-        txtptr+=2;
-        g_free(buff);
-        continue;
-
-        /* if it's a color read the color, set the current color and
-           continue */
-      } else if (!strcasecmp(buff, "@color") 
-                 && owl_global_is_colorztext(&g)) {
-        g_free(buff);
-        txtptr+=7;
-        tmpptr=strpbrk(txtptr, "@{[<()>]}");
-        if (tmpptr &&
-            ((txtptr[-1]=='(' && tmpptr[0]==')') ||
-             (txtptr[-1]=='<' && tmpptr[0]=='>') ||
-             (txtptr[-1]=='[' && tmpptr[0]==']') ||
-             (txtptr[-1]=='{' && tmpptr[0]=='}'))) {
-
-          /* grab the color name */
-          buff=g_new(char, tmpptr-txtptr+20);
-          strncpy(buff, txtptr, tmpptr-txtptr);
-          buff[tmpptr-txtptr]='\0';
-
-          /* set it as the current color */
-          curcolor=owl_util_string_to_color(buff);
-          if (curcolor == OWL_COLOR_INVALID)
-              curcolor = OWL_COLOR_DEFAULT;
-          g_free(buff);
-          txtptr=tmpptr+1;
-          continue;
-
-        } else {
-
-        }
-
-      } else {
-        /* if we didn't understand it, we'll print it.  This is different from zwgc
-         * but zwgc seems to be smarter about some screw cases than I am
-         */
-        g_free(buff);
-        owl_fmtext_append_attr(f, "@", curattrs, curcolor, OWL_COLOR_DEFAULT);
-        txtptr++;
-        continue;
-      }
-
-    } else if (ptr[0]=='}' || ptr[0]==']' || ptr[0]==')' || ptr[0]=='>') {
-      /* add the text up to this point first */
-      buff=g_new(char, ptr-txtptr+20);
-      strncpy(buff, txtptr, ptr-txtptr);
-      buff[ptr-txtptr]='\0';
-      owl_fmtext_append_attr(f, buff, curattrs, curcolor, OWL_COLOR_DEFAULT);
-      g_free(buff);
-
-      /* now deal with the closer */
-      txtptr=ptr;
-
-      /* first, if the stack is empty we must bail (just print and go) */
-      if (stacksize==0) {
-        buff=g_new(char, 5);
-        buff[0]=ptr[0];
-        buff[1]='\0';
-        owl_fmtext_append_attr(f, buff, curattrs, curcolor, OWL_COLOR_DEFAULT);
-        g_free(buff);
-        txtptr++;
-        continue;
-      }
-
-      /* if the closing char is what's on the stack, turn off the
-         attribue and pop the stack */
-      if ((ptr[0]==')' && chrstack[stacksize-1]=='(') ||
-          (ptr[0]=='>' && chrstack[stacksize-1]=='<') ||
-          (ptr[0]==']' && chrstack[stacksize-1]=='[') ||
-          (ptr[0]=='}' && chrstack[stacksize-1]=='{')) {
-        int i;
-        stacksize--;
-        curattrs=OWL_FMTEXT_ATTR_NONE;
-        curcolor = colorstack[stacksize];
-        for (i=0; i<stacksize; i++) {
-          curattrs|=attrstack[i];
-        }
-        txtptr+=1;
-        continue;
-      } else {
-        /* otherwise print and continue */
-        buff=g_new(char, 5);
-        buff[0]=ptr[0];
-        buff[1]='\0';
-        owl_fmtext_append_attr(f, buff, curattrs, curcolor, OWL_COLOR_DEFAULT);
-        g_free(buff);
-        txtptr++;
-        continue;
-      }
-    } else {
-      /* we've found an unattached opener, print everything and move on */
-      buff=g_new(char, ptr-txtptr+20);
-      strncpy(buff, txtptr, ptr-txtptr+1);
-      buff[ptr-txtptr+1]='\0';
-      owl_fmtext_append_attr(f, buff, curattrs, curcolor, OWL_COLOR_DEFAULT);
-      g_free(buff);
-      txtptr=ptr+1;
-      continue;
-    }
-  }
-}
+  ztext_env *tree = ztext_tree(text);
+
+  owl_fmtext_append_ztext_tree(f, tree,
+                               OWL_FMTEXT_ATTR_NONE, OWL_COLOR_DEFAULT);
+
+  ztext_env_free(tree);
+}
+
 
 /* requires that the list values are strings or NULL.

owl.h

@@ -63 +63 @@
 
 #include "window.h"
+#include "ztext.h"
 
 extern const char *version;

tester.c

@@ -26 +26 @@
 int call_filter_regtest(void);
 int owl_smartstrip_regtest(void);
+int ztext_test(void);
 
 extern void owl_perl_xs_init(pTHX);
@@ -118 +119 @@
   numfailures += call_filter_regtest();
   numfailures += owl_smartstrip_regtest();
+  numfailures += ztext_test();
   if (numfailures) {
       fprintf(stderr, "# *** WARNING: %d failures total\n", numfailures);
@@ -1055 +1057 @@
   return numfailed;
 }
+
+static char *ztext_str(ztext_env *tree)
+{
+    /* Unambiguous representation of the parse tree */
+    char *s, *t, *u;
+
+    if (tree->opener) {
+        s = g_strdup_printf("{'%s%c'", tree->label, tree->opener);
+    } else {
+        s = g_strdup("{-");
+    }
+
+    for (ztext_node *p = tree->content; p != NULL; p = p->next) {
+        t = s;
+        if (p->type== ZTEXT_NODE_STRING) {
+            s = g_strdup_printf("%s |%s|", s, p->string);
+        } else if (p->type == ZTEXT_NODE_ENV) {
+            u = ztext_str(p->env);
+            s = g_strconcat(s, " ", u, NULL);
+            g_free(u);
+        }
+        g_free(t);
+    }
+
+    t = s;
+    if (tree->closer) {
+        s = g_strdup_printf("%s '%c'}", s, tree->closer);
+    } else {
+        s = g_strconcat(s, "}", NULL);
+    }
+    g_free(t);
+
+    return s;
+}
+
+char *ztext_ztext(ztext_env *tree)
+{
+    char *s, *t, *u;
+
+    if (tree->closer == '@' && tree->content == NULL)
+        return g_strdup("@@");
+
+    if (tree->opener) {
+        s = g_strdup_printf("%s%c", tree->label, tree->opener);
+    } else {
+        s = g_strdup("");
+    }
+
+    for (ztext_node *p = tree->content; p != NULL; p = p->next) {
+        t = s;
+        if (p->type == ZTEXT_NODE_STRING) {
+            s = g_strconcat(s, p->string, NULL);
+        } else if (p->type == ZTEXT_NODE_ENV) {
+            u = ztext_ztext(p->env);
+            s = g_strconcat(s, u, NULL);
+            g_free(u);
+        }
+        g_free(t);
+    }
+
+    if (tree->closer) {
+        t = s;
+        s = g_strdup_printf("%s%c", s, tree->closer);
+        g_free(t);
+    }
+
+    return s;
+}
+
+int ztext_test(void)
+{
+    int numfailed = 0;
+    char *s;
+    ztext_env *t;
+    char *d;
+
+#define CHECK_ZTEXT_STR(in, expected)                                         \
+    do {                                                                      \
+      t = ztext_tree(in);                                                     \
+      s = ztext_str(t);                                                       \
+      d = g_strdup_printf("ztext decode \"%s\" expected \"%s\" got \"%s\"",   \
+                          in, expected, s);                                   \
+      FAIL_UNLESS(d, !strcmp(s, expected));                                   \
+      ztext_env_free(t);                                                      \
+      g_free(d);                                                              \
+      g_free(s);                                                              \
+    } while (0)
+
+    CHECK_ZTEXT_STR("", "{-}");
+    CHECK_ZTEXT_STR("foo", "{- |foo|}");
+    CHECK_ZTEXT_STR("@{foo}", "{- {'@{' |foo| '}'}}");
+    CHECK_ZTEXT_STR("@bar{foo}", "{- {'@bar{' |foo| '}'}}");
+    CHECK_ZTEXT_STR("@bar{foo@bar}", "{- {'@bar{' |foo@bar| '}'}}");
+    CHECK_ZTEXT_STR("@bar{foo@@bar}", "{- {'@bar{' |foo| {'@@' '@'} |bar| '}'}}");
+    CHECK_ZTEXT_STR("@{foo@}bar}baz", "{- {'@{' |foo@| '}'} |bar}baz|}");
+    CHECK_ZTEXT_STR("foo@bar{baz@(bang})}",
+                    "{- |foo| {'@bar{' |baz| {'@(' |bang}| ')'} '}'}}");
+    CHECK_ZTEXT_STR("foo@bar{baz}", "{- |foo| {'@bar{' |baz| '}'}}");
+    CHECK_ZTEXT_STR("@bloop", "{- |@bloop|}");
+
+#define CHECK_ZTEXT_STRIP(in, expected)                                      \
+    do {                                                                     \
+      t = ztext_tree(in);                                                    \
+      s = ztext_strip(t);                                                    \
+      d = g_strdup_printf("ztext strip \"%s\" expected \"%s\" got \"%s\"",   \
+                          in, expected, s);                                  \
+      FAIL_UNLESS(d, !strcmp(s, expected));                                  \
+      ztext_env_free(t);                                                     \
+      g_free(d);                                                             \
+      g_free(s);                                                             \
+    } while (0)
+
+    CHECK_ZTEXT_STRIP("", "");
+    CHECK_ZTEXT_STRIP("foo", "foo");
+    CHECK_ZTEXT_STRIP("@{foo}", "foo");
+    CHECK_ZTEXT_STRIP("@bar{foo}", "foo");
+    CHECK_ZTEXT_STRIP("@bar{foo@bar}", "foo@bar");
+    CHECK_ZTEXT_STRIP("@bar{foo@@bar}", "foo@bar");
+    CHECK_ZTEXT_STRIP("@{foo@}bar}baz", "foo@bar}baz");
+    CHECK_ZTEXT_STRIP("foo@bar{baz@(bang})}", "foobazbang}");
+    CHECK_ZTEXT_STRIP("foo@bar{baz}", "foobaz");
+    CHECK_ZTEXT_STRIP("@bloop", "@bloop");
+
+#define CHECK_ZTEXT_RT(in)                                                \
+    do {                                                                  \
+      t = ztext_tree(in);                                                 \
+      s = ztext_ztext(t);                                                 \
+      d = g_strdup_printf("ztext round-trip \"%s\" got \"%s\"", in, s);   \
+      FAIL_UNLESS(d, !strcmp(s, in));                                     \
+      ztext_env_free(t);                                                  \
+      g_free(s);                                                          \
+    } while (0)
+
+    CHECK_ZTEXT_RT("");
+    CHECK_ZTEXT_RT("foo");
+    CHECK_ZTEXT_RT("@{foo}");
+    CHECK_ZTEXT_RT("@bar{foo}");
+    CHECK_ZTEXT_RT("@bar{foo@bar}");
+    CHECK_ZTEXT_RT("@bar{foo@@bar}");
+    CHECK_ZTEXT_RT("@{foo@}bar}baz");
+    CHECK_ZTEXT_RT("foo@bar{baz@(bang})}");
+    CHECK_ZTEXT_RT("foo@bar{baz}");
+    CHECK_ZTEXT_RT("@bloop");
+
+    /* fuzzzzzz */
+    srand48(getpid() + time(0));
+
+    char tango[32];
+    const char CHARS[] = "@abc ({<[]>})";
+
+    for(int i=0; i < 1000; i++) {
+        memset(tango, 0, sizeof(tango));
+        int l = lrand48() % sizeof(tango);
+        for (int j=0; j < l; j++) {
+            tango[j] = CHARS[lrand48() % (sizeof(CHARS) -1)];
+        }
+        CHECK_ZTEXT_RT(tango);
+    }
+
+
+    return numfailed;
+}