compare against git diff gzipped

this is not looking good...

2 files changed, 70 insertions, 21 deletions

diff --git a/exp/Makefile b/exp/Makefile
index e7314bf..16d4b65 100644
--- a/exp/Makefile
+++ b/exp/Makefile
@@ -4,23 +4,50 @@ REPO_PATH := repo
 GIT_PATH  := git
 GITC      := git -C $(REPO_PATH)
 
-TEMP      := temp
+MAX_VERSION ?= 3
+
+DATADIRS  := backup diffs
+SIZEFILES := $(DATADIRS:%=%-size)
 
 .PHONY: all exp
-all exp: versions results
-	@echo "total size    : `head --lines=1 results | cut -f1`"
-	@echo "version count : `wc -l versions`"
-	@echo "repo results  :\n`cat results`"
+all exp: summary.csv $(SIZEFILES)
+	@echo "============== SUMMARY =============="
+	@cat $< | tr ',' '\t'
+	@echo "==============  TOTAL  =============="
+	@for i in $(SIZEFILES); do cat $$i | paste -sd+ | bc; done | tr '\n' '\t'
+	@echo
+
+summary.csv: $(SIZEFILES)
+	echo $(DATADIRS) | tr ' ' ',' > $@
+	paste -d "," $^ >> $@
 
-versions: results
-	grep [0-9]$$ $< | cut -f1 > $@
+backup-size: versions
+	rm -rf $@
+	for i in $</*; do \
+		cut -f1 $$i | paste -sd+ | bc >> $@; \
+	done
 
-results: | $(TEMP)
-	du -bad 2 $(TEMP) | sort -k2 > $@
+diffs-size: diffs
+	find $< -type f -exec du -ba {} + | cut -f1 > $@
 
-$(TEMP): commits ../dna-backup
+versions: results backup
 	rm -rf $@
-	./exp.sh $< $(REPO_PATH) $@
+	mkdir -p $@
+	for i in backup/*; do \
+		v=`echo $$i | cut -d / -f2`; \
+		grep $$i $< > $@/$$v ; \
+	done
+
+results: backup
+	find $< -type f -exec du -ba {} + \
+	| grep -v hashes \
+	| sort -k2 \
+	> $@
+
+$(DATADIRS): commits ../dna-backup
+	rm -rf $(DATADIRS)
+	./exp.sh $< $(REPO_PATH) $(MAX_VERSION) $(DATADIRS)
+	touch $(DATADIRS)
 
 ../dna-backup: .FORCE
 	@$(MAKE) -C .. --no-print-directory dna-backup
@@ -36,8 +63,12 @@ repo git:
 	$(GITC) config advice.detachedHead false
 
 .PHONY: clean
-clean:
-	rm -rf $(REPO_PATH) $(GIT_PATH) $(TEMP)
-	rm -f commits results versions
+clean: mostlyclean
+	rm -rf $(REPO_PATH) $(GIT_PATH)
+	rm commits
+
+mostlyclean:
+	rm -rf $(DATADIRS) versions
+	rm -f results summary.csv $(SIZEFILES)
 
 .FORCE: ;
diff --git a/exp/exp.sh b/exp/exp.sh
index 8c861a3..0213b2f 100755
--- a/exp/exp.sh
+++ b/exp/exp.sh
@@ -1,12 +1,30 @@
 #!/bin/bash
 
-commits_file=$1
-repo_path=$2
-temp=$3
+commits="$1"
+repo="$2"
+max_count="$3"
+backup="$4"
+diffs="$5"
 
-cat $commits_file | while read i
+mkdir -p $backup $diffs
+
+# "empty tree" commit
+prev="4b825dc642cb6eb9a060e54bf8d69288fbee4904"
+last=$(tail --lines=1 "$commits" | cut -f1)
+
+i=0
+cat "$commits" | while read line
 do
-	hash=$(echo "$i" | cut -f1)
-	git -C $repo_path checkout $hash
-	../dna-backup commit -v 2 $repo_path $temp
+	hash=$(echo "$line" | cut -f1)
+	git -C "$repo" checkout "$hash"
+	git -C "$repo" diff --minimal --binary --unified=0 "$prev" | gzip > "$diffs/$i.diff.gz"
+	../dna-backup commit -v 2 "$repo" "$backup"
+	prev="$hash"
+	let i++
+	if [[ $i == $max_count ]]
+	then
+		break
+	fi
 done
+
+git -C "$repo" checkout "$last"