From 044392aed99af6868a1e94251feb03ae89c55f8c Mon Sep 17 00:00:00 2001
From: "Eric S. Raymond" <esr@thyrsus.com>
Date: Wed, 17 Sep 2003 16:44:39 +0000
Subject: [PATCH] Better statistics in the count fields.

---
 main.c           | 18 +++++++++++-----
 report.c         | 54 ++++++++++++++++++++++++++++++++++++++++++++----
 scf-standard.xml | 15 ++++++++++++--
 shred.h          |  5 ++++-
 4 files changed, 80 insertions(+), 12 deletions(-)

diff --git a/main.c b/main.c
index c35d29d..83bd83c 100644
--- a/main.c
+++ b/main.c
@@ -601,15 +601,23 @@ main(int argc, char *argv[])
     puts("Merge-Program: comparator " VERSION);
     printf("Normalization: %s\n", scflist->normalization);
     printf("Shred-Size: %d\n", scflist->shred_size);
-    puts("%%");
-    for (scf = scflist; scf->next; scf = scf->next)
-	printf("%s:%d\n", scf->name, scf->totallines);
-    puts("%%");
 
     report_time("Hash merge done, %d shreds", sort_count);
     sort_hashes(sort_buffer, sort_count);
     report_time("Sort done");
-    emit_report(sort_buffer, sort_count);
+
+    emit_report1(sort_buffer, sort_count);
+
+    puts("%%");
+    for (scf = scflist; scf->next; scf = scf->next)
+	printf("%s:%d:%d:%d\n", 
+	       scf->name, 
+	       match_count(scf->name), 
+	       line_count(scf->name), 
+	       scf->totallines);
+    puts("%%");
+
+    emit_report2();
 
     exit(0);
 }
diff --git a/report.c b/report.c
index 15d4220..763c284 100644
--- a/report.c
+++ b/report.c
@@ -339,11 +339,13 @@ static int sortmatch(const void *a, const void *b)
     return(0);
 }
 
-void emit_report(struct sorthash_t *obarray, int hashcount)
-/* report our results */
+static struct match_t *hitlist;
+static int mergecount;
+
+void emit_report1(struct sorthash_t *obarray, int hashcount)
+/* report our results (header portion) */
 {
-    struct match_t *hitlist, *match, *copy;
-    int mergecount;
+    struct match_t *match, *copy;
 
     hitlist = reduce_matches(obarray, &hashcount);
     if (debug)
@@ -360,6 +362,15 @@ void emit_report(struct sorthash_t *obarray, int hashcount)
 	if (match->nmatches > 0 && copy < match)
 	    *copy++ = *match;
     mergecount = (copy - hitlist);
+
+    printf("Matches: %d\n", mergecount);
+ }
+
+void emit_report2(void)
+/* report our results (matches) */
+{
+    struct match_t *match;
+
     qsort(hitlist, mergecount, sizeof(struct match_t), sortmatch);
 
     for (match = hitlist; match < hitlist + mergecount; match++)
@@ -397,4 +408,39 @@ void emit_report(struct sorthash_t *obarray, int hashcount)
     /* free(hitlist); */
 }
 
+int match_count(const char *name)
+/* return count of matches with given tree in them */
+{
+    struct match_t *match;
+    int i, count = 0;
+
+    for (match = hitlist; match < hitlist + mergecount; match++)
+	for (i=0; i < match->nmatches; i++)
+	    if (strncmp(name, match->matches[i].file->name, strlen(name)))
+	    {
+		count++;
+		break;
+	    }
+
+    return(count);
+}
+
+int line_count(const char *name)
+/* return number of lines from given tree in matching segments */
+{
+    struct match_t *match;
+    int i, count = 0;
+
+    for (match = hitlist; match < hitlist + mergecount; match++)
+	for (i=0; i < match->nmatches; i++)
+	    if (strncmp(name, match->matches[i].file->name, strlen(name)))
+	    {
+		count += match->matches[i].hash.end -  match->matches[i].hash.start + 1;
+		break;
+	    }
+
+    return(count);
+}
+
+
 /* report.c ends here */
diff --git a/scf-standard.xml b/scf-standard.xml
index c5f394c..56dfad2 100644
--- a/scf-standard.xml
+++ b/scf-standard.xml
@@ -344,9 +344,20 @@ comparisons.</para>
 </variablelist>
 
 <para>The data section consists of a statistics block followed by a
-sequence of range sets.  The statistics block consists of ant number 
+sequence of range sets.  The statistics block consists of any number 
 of count lines followed by a %%\n section marker. Each linecount line
-consists of two colon-separated fields, a tree name and a line count.</para>
+consists of four colon-separated fields, 
+
+<orderedlist>
+<listitem><para>The name of a tree</para></listitem>
+<listitem><para>A count of chunks containing files from the tree</para></listitem>
+<listitem><para>A count of common lines from the tree</para></listitem>
+<listitem><para>The total line count of the tree</para></listitem>
+</orderedlist>
+
+
+
+a tree name and a line count.</para>
 
 <para>Each range set consists of two or more ranges followed by a line
 containing a %%\n section marker.  A range is represented as four
diff --git a/shred.h b/shred.h
index ecd9a33..33d6a7a 100644
--- a/shred.h
+++ b/shred.h
@@ -67,6 +67,9 @@ extern int shredfile(struct filehdr_t *,
 extern void sort_hashes(struct sorthash_t *hashlist, int hashcount);
 
 /* shredcompare.c functions */
-extern void emit_report(struct sorthash_t *obarray, int hashcount);
+extern void emit_report1(struct sorthash_t *obarray, int hashcount);
+extern void emit_report2(void);
+extern int match_count(const char *name);
+extern int line_count(const char *name);
 
 /* shred.h ends here */
-- 
2.43.0