Changeset 39217

Timestamp:

Dec 3, 2015, 4:06:46 PM (11 years ago)

Author:

eugene

Message:

handle the case of a file without a <return> on the last line

File:

• trunk/Ohana/src/opihi/cmd.data/read_vectors.c (modified) (6 diffs)

trunk/Ohana/src/opihi/cmd.data/read_vectors.c

@@ -41 +41 @@
   time_t TimeReference;
   int i, j, Nskip, Narg, IsCSV, VERBOSE;
-  int Nbytes, Nstart, NELEM, Nelem, nread;
+  int Nbytes, NELEM, nread;
   char *colstr, *c0, *c1, *extname;
   char varname[1024];  // used as a buffer for the names of string fields
@@ -188 +188 @@
   }
   
+  // we allocate one extra byte into which we never read so there will always be a NULL terminating the string
   ALLOCATE (buffer, char, 0x10001);
   bzero (buffer, 0x10001);
-
-  /* 
-  for (i = 0; i < Nskip; i++) {
-    if (scan_line_maxlen (f, buffer, 0x10000) == EOF) {
-      gprint (GP_ERR, "problem reading file %s\n", filename);
-      read_vectors_cleanup();
-      return FALSE;
-    }
-  }
-  */
 
   int Nline_read = 0; // track number of lines read so far (use to skip lines as well)
@@ -209 +200 @@
   // we treat \n\r pair as a single EOL char to handle mac files:
 
-  Nstart = 0; // location of the last valid byte in the buffer (start filling here)
-  Nelem = 0; // number of valid rows read (vector elements)
-  while (TRUE) {
+  int Nstart = 0; // location of the last valid byte in the buffer (start filling here)
+  int Nelem = 0; // number of valid rows read (vector elements)
+  int EndOfFile = FALSE;
+  while (!EndOfFile) {
     Nbytes = 0x10000 - Nstart;
-    bzero (&buffer[Nstart], Nbytes);
+    // we have allocated one extra byte into which we never read so there will always be a NULL terminating the string
+    bzero (&buffer[Nstart], Nbytes + 1);
     nread = fread (&buffer[Nstart], 1, Nbytes, f);
     if (ferror (f)) {
@@ -219 +212 @@
       break;
     }
-    if (nread == 0) break; // end of the file
-    // nbytes = nread + Nstart;
+    // we still need to parse the rest of the buffer, but there might not be an EOL on the last line
+    if (nread == 0) {
+      EndOfFile = TRUE;
+    }
     
     int bufferStatus = TRUE; 
@@ -227 +222 @@
       c1 = strchr (c0, '\n'); // find the end of this current line (also valid for a Mac: \r\n)
       if (!c1) {
-        c1 = strchr (c0, '\r'); // try \r for non-UNIX files
-      }
-      if (c1 == (char *) NULL) {
+        c1 = strchr (c0, '\r'); // try \r for Windows files
+      }
+      if (!c1) {
         Nstart = strlen (c0);
-        memmove (buffer, c0, Nstart);
-        bufferStatus = FALSE;
-        continue;
+        if (EndOfFile) {
+          // if we have reached EOF, we need to do one last pass in case there is a line without a return
+          c1 = c0 + Nstart;
+          bufferStatus = FALSE;
+        } else {
+          // if we have not reached EOF, we need to shift the buffer to the start of this line and read more data
+          memmove (buffer, c0, Nstart);
+          bufferStatus = FALSE;
+          continue;
+        }
       }
       *c1 = 0; // mark the end of the line 
       Nline_read ++;
 
-      if (Nline_read <= Nskip) { c0 = c1 + 1; continue; }
-
-      if (*c0 == '#') { c0 = c1 + 1; continue; }
-      if (*c0 == '!') { c0 = c1 + 1; continue; }
+      // skip to the next line (but if EOF, do not overrun buffer)
+      if (Nline_read <= Nskip) { if (!EndOfFile) { c0 = c1 + 1; } continue; }
+      if (*c0 == '#')          { if (!EndOfFile) { c0 = c1 + 1; } continue; }
+      if (*c0 == '!')          { if (!EndOfFile) { c0 = c1 + 1; } continue; }
 
       // parse the vectors in this line.  this code is a bit inefficient: each column
@@ -319 +321 @@
         }
       }
-      c0 = c1 + 1;
+      if (!EndOfFile) {
+        c0 = c1 + 1;
+      }
     }
   }