Changeset 4f70ebc in git

Timestamp:

14/01/14 02:23:57 (6 years ago)

Author:

Olly Betts <olly@…>

Branches:

line_contents, master, stereo, travis-osx

Children:

5dc0378

Parents:

ad95991

git-author:

Olly Betts <olly@…> (14/01/14 00:44:01)

git-committer:

Olly Betts <olly@…> (14/01/14 02:23:57)

Message:

src/: Add new "datestamp_numeric" field to struct img giving the
datestamp as a time_t in UTC (or (time_t)-1 if there's no datestamp
or we failed to convert it). For .3d >= v8, this field is reliable.
We attempt to convert date strings in .3d <= v7 and CMAP XYZ
files, but may get the timezone wrong.

Files:

• ChangeLog (modified) (1 diff)
• src/dump3d.c (modified) (2 diffs)
• src/img.c (modified) (5 diffs)
• src/img.h (modified) (2 diffs)
• src/mainfrm.cc (modified) (2 diffs)

ChangeLog

@@ +1 @@
+Tue Jan 14 00:40:06 GMT 2014  Olly Betts <olly@survex.com>
+
+        * src/: Add new "datestamp_numeric" field to struct img giving the
+          datestamp as a time_t in UTC (or (time_t)-1 if there's no datestamp
+          or we failed to convert it).  For .3d >= v8, this field is reliable.
+          We attempt to convert date strings in .3d <= v7 and CMAP XYZ
+          files, but may get the timezone wrong.
+
 Tue Jan 14 00:29:35 GMT 2014  Olly Betts <olly@survex.com>
 

src/dump3d.c

@@ -1 +1 @@
 /* dump3d.c */
 /* Show raw contents of .3d file in text form */
-/* Copyright (C) 2001,2002,2006,2011,2012,2013 Olly Betts
+/* Copyright (C) 2001,2002,2006,2011,2012,2013,2014 Olly Betts
  *
  * This program is free software; you can redistribute it and/or modify
@@ -94 +94 @@
    printf("TITLE \"%s\"\n", pimg->title);
    printf("DATE \"%s\"\n", pimg->datestamp);
+   printf("DATE_NUMERIC %ld\n", pimg->datestamp_numeric);
    printf("VERSION %d\n", pimg->version);
    if (pimg->is_extended_elevation)

src/img.c

@@ -22 +22 @@
 #endif
 
+#include <ctype.h>
+#include <errno.h>
+#include <limits.h>
+#include <locale.h>
 #include <stdio.h>
+#include <stdlib.h>
 #include <string.h>
-#include <stdlib.h>
 #include <time.h>
-#include <ctype.h>
-#include <locale.h>
 
 #include "img.h"
@@ -152 +154 @@
 #endif
 
+static char * my_strdup(const char *str);
+
+static time_t
+mktime_with_tz(struct tm * tm, const char * tz)
+{
+    time_t r;
+    char * old_tz = getenv("TZ");
+#ifdef _WIN32
+    if (old_tz) {
+        old_tz = my_strdup(old_tz);
+        if (!old_tz)
+            return (time_t)-1;
+    }
+    if (_putenv_s("TZ", tz) != 0) {
+        osfree(old_tz);
+        return (time_t)-1;
+    }
+#elif defined HAVE_SETENV
+    if (old_tz) {
+        old_tz = my_strdup(old_tz);
+        if (!old_tz)
+            return (time_t)-1;
+    }
+    if (setenv("TZ", tz, 1) < 0) {
+        osfree(old_tz);
+        return (time_t)-1;
+    }
+#else
+    char * p;
+    if (old_tz) {
+        size_t len = strlen(old_tz) + 1;
+        p = (char *)xosmalloc(len + 3);
+        if (!p)
+            return (time_t)-1;
+        memcpy(p, "TZ=", 3);
+        memcpy(p + 3, tz, len);
+        old_tz = p;
+    }
+    p = (char *)xosmalloc(strlen(tz) + 4);
+    if (!p) {
+        osfree(old_tz);
+        return (time_t)-1;
+    }
+    memcpy(p, "TZ=", 3);
+    strcpy(p + 3, tz);
+    if (putenv(p) != 0) {
+        osfree(p);
+        osfree(old_tz);
+        return (time_t)-1;
+    }
+#endif
+    tzset();
+    r = mktime(tm);
+    if (old_tz) {
+#ifdef _WIN32
+        _putenv_s("TZ", old_tz);
+#elif !defined HAVE_SETENV
+        putenv(old_tz);
+        osfree(p);
+#else
+        setenv("TZ", old_tz, 1);
+#endif
+        osfree(old_tz);
+    } else {
+#ifdef _WIN32
+        _putenv_s("TZ", "");
+#elif !defined HAVE_UNSETENV
+        putenv((char*)"TZ");
+        osfree(p);
+#else
+        unsetenv("TZ");
+#endif
+    }
+    return r;
+}
+
 static unsigned short
 getu16(FILE *fh)
@@ -348 +426 @@
 
    pimg->title = pimg->datestamp = NULL;
+   pimg->datestamp_numeric = (time_t)-1;
+
    if (survey) {
       len = strlen(survey);
@@ -481 +561 @@
          goto error;
       }
-      /* FIXME: reparse date? */
+      /* There doesn't seem to be a spec for what happens after 1999 with cmap
+       * files, so this code allows for:
+       *  * 21xx -> xx (up to 2150)
+       *  * 21xx -> 1xx (up to 2199)
+       *  * full year being specified instead of 2 digits
+       */
       len = strlen(line);
-      if (len > 59) line[59] = '\0';
+      if (len > 59) {
+         /* Don't just truncate at column 59, allow for a > 2 digit year. */
+         char * p = strstr(line + len, "Page");
+         if (p) {
+            while (p > line && p[-1] == ' ')
+               --p;
+            *p = '\0';
+            len = p - line;
+         } else {
+            line[59] = '\0';
+         }
+      }
       if (len > 45) {
+         /* YY/MM/DD HH:MM */
+         struct tm tm;
+         unsigned long v;
+         char * p;
          pimg->datestamp = my_strdup(line + 45);
+         p = pimg->datestamp;
+         v = strtoul(p, &p, 10);
+         if (v <= 50) {
+            /* In the absence of a spec for cmap files, assume <= 50 means 21st
+             * century. */
+            v += 2000;
+         } else if (v < 200) {
+            /* Map 100-199 to 21st century. */
+            v += 1900;
+         }
+         if (v == ULONG_MAX || *p++ != '/')
+            goto bad_cmap_date;
+         tm.tm_year = v - 1900;
+         v = strtoul(p, &p, 10);
+         if (v < 1 || v > 12 || *p++ != '/')
+            goto bad_cmap_date;
+         tm.tm_mon = v - 1;
+         v = strtoul(p, &p, 10);
+         if (v < 1 || v > 31 || *p++ != ' ')
+            goto bad_cmap_date;
+         tm.tm_mday = v;
+         v = strtoul(p, &p, 10);
+         if (v >= 24 || *p++ != ':')
+            goto bad_cmap_date;
+         tm.tm_hour = v;
+         v = strtoul(p, &p, 10);
+         if (v >= 60)
+            goto bad_cmap_date;
+         tm.tm_min = v;
+         if (*p == ':') {
+            v = strtoul(p + 1, &p, 10);
+            if (v > 60)
+               goto bad_cmap_date;
+            tm.tm_sec = v;
+         } else {
+            tm.tm_sec = 0;
+         }
+         tm.tm_isdst = 0;
+         /* We have no indication of what timezone this timestamp is in.  It's
+          * probably local time for whoever processed the data, so just assume
+          * UTC, which is at least fairly central in the possibilities.
+          */
+         pimg->datestamp_numeric = mktime_with_tz(&tm, "");
       } else {
          pimg->datestamp = my_strdup(TIMENA);
       }
+bad_cmap_date:
       if (strncmp(line, "  Cave Survey Data Processed by CMAP ",
                   LITLEN("  Cave Survey Data Processed by CMAP ")) == 0) {
@@ -622 +766 @@
       }
    }
+
+   if (pimg->datestamp[0] == '@') {
+      unsigned long v;
+      char * p;
+      errno = 0;
+      v = strtoul(pimg->datestamp + 1, &p, 10);
+      if (errno == 0 && *p == '\0')
+         pimg->datestamp_numeric = v;
+      /* FIXME: We're assuming here that the C time_t epoch is 1970, which is
+       * true for Unix-like systems, Mac OS X and Windows, but isn't guaranteed
+       * by ISO C.
+       */
+   } else {
+      /* %a,%Y.%m.%d %H:%M:%S %Z */
+      struct tm tm;
+      unsigned long v;
+      char * p = pimg->datestamp;
+      while (isalpha((unsigned char)*p)) ++p;
+      if (*p == ',') ++p;
+      while (isspace((unsigned char)*p)) ++p;
+      v = strtoul(p, &p, 10);
+      if (v == ULONG_MAX || *p++ != '.')
+         goto bad_3d_date;
+      tm.tm_year = v - 1900;
+      v = strtoul(p, &p, 10);
+      if (v < 1 || v > 12 || *p++ != '.')
+         goto bad_3d_date;
+      tm.tm_mon = v - 1;
+      v = strtoul(p, &p, 10);
+      if (v < 1 || v > 31 || *p++ != ' ')
+         goto bad_3d_date;
+      tm.tm_mday = v;
+      v = strtoul(p, &p, 10);
+      if (v >= 24 || *p++ != ':')
+         goto bad_3d_date;
+      tm.tm_hour = v;
+      v = strtoul(p, &p, 10);
+      if (v >= 60 || *p++ != ':')
+         goto bad_3d_date;
+      tm.tm_min = v;
+      v = strtoul(p, &p, 10);
+      if (v > 60)
+         goto bad_3d_date;
+      tm.tm_sec = v;
+      tm.tm_isdst = 0;
+      while (isspace((unsigned char)*p)) ++p;
+      /* p now points to the timezone string.
+       *
+       * However, it's likely to be a string like "BST", and such strings can
+       * be ambiguous (BST could be UTC+1 or UTC+6), so it is impossible to
+       * reliably convert in all cases.  Just pass what we have to tzset() - if
+       * it doesn't handle it, UTC will be used.
+       */
+      pimg->datestamp_numeric = mktime_with_tz(&tm, p);
+   }
+bad_3d_date:
 
    pimg->start = ftell(pimg->fh);

src/img.h

@@ -1 +1 @@
 /* img.h
  * Header file for routines to read and write Survex ".3d" image files
- * Copyright (C) Olly Betts 1993,1994,1997,2001,2002,2003,2004,2005,2006,2010,2011,2012,2013
+ * Copyright (C) Olly Betts 1993,1994,1997,2001,2002,2003,2004,2005,2006,2010,2011,2012,2013,2014
  *
  * This program is free software; you can redistribute it and/or modify
@@ -99 +99 @@
     */
    char *datestamp;
+   /* The datestamp as a time_t (or (time_t)-1 if not available).
+    *
+    * For 3d format versions >= 8, this is a reliable value and in UTC.  Older
+    * 3d format versions store a human readable time, which img will attempt
+    * to decode, but it may fail, particularly with handling timezones.  Even
+    * if it does work, beware that times in local time where DST applies are
+    * inherently ambiguous around when the clocks go back.
+    *
+    * CMAP XYZ files contain a timestamp.  It's probably in localtime (but
+    * without any timezone information) and the example files are all pre-2000
+    * and have two digit years.  We do our best to turn these into a useful
+    * time_t value.
+    */
+   time_t datestamp_numeric;
    char separator; /* character used to separate survey levels ('.' usually) */
 #if IMG_API_VERSION == 0

src/mainfrm.cc

@@ -5 +5 @@
 //
 //  Copyright (C) 2000-2002,2005,2006 Mark R. Shinwell
-//  Copyright (C) 2001-2003,2004,2005,2006,2010,2011,2012,2013 Olly Betts
+//  Copyright (C) 2001-2003,2004,2005,2006,2010,2011,2012,2013,2014 Olly Betts
 //  Copyright (C) 2005 Martin Green
 //
@@ -1338 +1338 @@
         m_DateStamp = wmsg(/*Date and time not available.*/108);
     } else if (survey->datestamp[0] == '@') {
-        char * end;
-        time_t t = (time_t)strtol(survey->datestamp + 1, &end, 10);
-        if (*end == '\0') {
-            const struct tm * tm = localtime(&t);
-            char buf[256];
-            strftime(buf, 256, msg(/*%a,%Y.%m.%d %H:%M:%S %Z*/107), tm);
-            m_DateStamp = wxString(buf, wxConvUTF8);
-        }
+        const struct tm * tm = localtime(&survey->datestamp_numeric);
+        char buf[256];
+        strftime(buf, 256, msg(/*%a,%Y.%m.%d %H:%M:%S %Z*/107), tm);
+        m_DateStamp = wxString(buf, wxConvUTF8);
     }
     if (m_DateStamp.empty()) {