[Fix #507] Period and duration parsers understand 0 units

tidyverse · Feb 7, 2017 · d515020 · d515020
1 parent f67a04c
commit d515020
Show file tree

Hide file tree

Showing 5 changed files with 24 additions and 17 deletions.
diff --git a/NEWS.md b/NEWS.md
@@ -10,6 +10,7 @@ Version 1.6.0.9000
 
 ### BUG FIXES
 
+* [#507](https://github.com/hadley/lubridate/issues/507) Period and duration parsers now understand 0 units.
 * [#466](https://github.com/hadley/lubridate/pull/466) Fix wrong formats within ymd_h family of functions.
 * [#472](https://github.com/hadley/lubridate/pull/472) Printing method for duration doesn't throw format error on fractional seconds.
 * [#475](https://github.com/hadley/lubridate/pull/475) character<> comparisons is no longer slow.

diff --git a/man/hms.Rd b/man/hms.Rd
diff --git a/src/period.c b/src/period.c
@@ -35,21 +35,22 @@ static const char *en_units[] = {"SECS", "secs", "seconds",
 #define N_PERIOD_UNITS 7
 
 intUnit parse_period_unit (const char **c) {
-  // S=0,  M=1, H=2, d=3, w=4, m=5, y=6
-  //  why this macro doesn't work?
-  /* SKIP_NON_ALPHANUMS(*c); */
+  // units: invalid=-1, S=0,  M=1, H=2, d=3, w=4, m=5, y=6
+  // SKIP_NON_ALPHANUMS(*c);   //  why this macro doesn't work here?
   while(**c && !(ALPHA(**c) || DIGIT(**c))) (*c)++;;
+
   intUnit out;
   out.unit = -1;
-
   out.val = parse_int(c, 100, FALSE);
-  if(out.val == 0) out.val = 1;
 
   if(**c){
     out.unit = parse_alphanum(c, en_units, N_EN_UNITS);
     if (out.unit < 0){
       return out;
     } else {
+      // if only unit name supplied, default to 1 units
+      if(out.val == -1)
+        out.val = 1;
       if(out.unit < 3) out.unit = 0;
       else if (out.unit < 6) out.unit = 1;
       else if (out.unit < 8) out.unit = 2;

diff --git a/src/tparse.c b/src/tparse.c
@@ -120,7 +120,9 @@ SEXP parse_dt(SEXP str, SEXP ord, SEXP formats, SEXP lt) {
             break;
           case 'y': // year in yy format
             y = parse_int(&c, 2, FALSE);
-			if (y <= 68)
+            if (y < 0)
+              succeed = 0;
+			else if (y <= 68)
 			  y += 2000;
 			else
 			  y += 1900;
@@ -132,9 +134,9 @@ SEXP parse_dt(SEXP str, SEXP ord, SEXP formats, SEXP lt) {
           case 'm': // month (allowing all months formats - m, b and B)
             SKIP_NON_ALPHANUMS(c);
             m = parse_int(&c, 2, FALSE);
-            if (m == 0) {
+            if (m == -1) { // failed
               m = parse_alpha_month(&c);
-              if (m == 0) {
+              if (m == 0) { // failed
                 SKIP_NON_DIGITS(c);
                 m = parse_int(&c, 2, FALSE);
               }
@@ -217,7 +219,7 @@ SEXP parse_dt(SEXP str, SEXP ord, SEXP formats, SEXP lt) {
             break;
           case 'z':
             // for %z: "+O100" or "+O1" or "+01:00"
-            if( !O_format ){
+            if( !O_format ) {
               if( !is_fmt ) {
                 while (*c && *c != '+' && *c != '-' && *c != 'Z') c++; // skip non + -
                 if( !*c ) { succeed = 0; break; };
@@ -226,21 +228,23 @@ SEXP parse_dt(SEXP str, SEXP ord, SEXP formats, SEXP lt) {
               if( *c == 'Z') {c++; break;}
               else if ( *c == '+' ) sig = -1;
               else if ( *c == '-') sig = 1;
-              else { succeed = 0; break; }
+              else {succeed = 0; break;}
               c++;
               Z = parse_int(&c, 2, FALSE);
+              if (Z < 0) {succeed = 0; break;}
               secs += sig*Z*3600;
               if( *c == ':' ){
                 c++;
-                if ( !DIGIT(*c) ){ succeed = 0; break; }
+                if ( !DIGIT(*c) ) {succeed = 0; break;}
               }
               if( DIGIT(*c) ){
                 Z = 0;
                 Z = parse_int(&c, 2, FALSE);
                 secs += sig*Z*60;
               }
               break;
-            } // else %Oz: "+0100". Pass through.
+            }
+            // else O_format %Oz: "+0100"; pass through
           case 'O':
             // %OO: "+01:00"
           case 'o':
@@ -253,14 +257,15 @@ SEXP parse_dt(SEXP str, SEXP ord, SEXP formats, SEXP lt) {
               else { succeed = 0; break; }
               c++;
               Z = parse_int(&c, 2, FALSE);
+              if (Z < 0) {succeed = 0; break;}
               secs += sig*Z*3600;
               if( *o == 'O'){
                 if ( *c == ':') c++;
 				else { succeed = 0; break; }
 			  }
-              if ( *o != 'o' ){
-                Z = 0;
+              if ( *o != 'o' ){ // z or O
                 Z = parse_int(&c, 2, FALSE);
+                if (Z < 0) {succeed = 0; break;}
                 secs += sig*Z*60;
               }
             } else error("Unrecognized format '%c' supplied", *o);

diff --git a/src/utils.c b/src/utils.c
@@ -47,14 +47,14 @@ int check_ymd(int y, int m, int d, int is_leap){
 /* parse N digit characters from **c. Return parsed non-negative integer. If
    failed to pass N chars, return -1.*/
 int parse_int (const char **c, const int N, const int strict) {
-  // maybe: fixme: this returns 0 if no parsing happened and strict = FALSE
   int tN = N, X = 0;
   while (DIGIT(**c) && tN > 0) {
     X = X * 10 + (**c - '0');
     (*c)++;
     tN--;
   }
-  if (strict && tN > 0) return -1;
+  if (strict && tN > 0) return -1; // not all numbers have been consumed
+  else if (tN == N) return -1; // no parsing happened
   else return X;
 }