fix: Update gocr tests to send PNGs instead of PNMs

test.mk

@@ -9,7 +9,6 @@
 
 PHONY: gocr__by_dpi
 gocr__by_dpi: ./runtime/bin/gocr.wasm.so
-	# cd ./tests/gocr/by_dpi && ./install.sh
 	cd ./tests/gocr/by_dpi  && ./run.sh
 
 PHONY: gocr__by_font

tests/gocr/by_dpi/.gitignore

@@ -1,4 +1,3 @@
-*.pnm
 *.png
 *.csv
 *.txt

tests/gocr/by_dpi/Makefile

@@ -16,22 +16,19 @@ client-72dpi:
 	mkdir -p tmp
 	shuf -n10 /usr/share/dict/american-english > tmp/10_words.txt
 	pango-view --dpi=72 --font=mono -qo tmp/72_dpi.png tmp/10_words.txt
-	pngtopnm tmp/72_dpi.png > tmp/72_dpi.pnm
-	curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary @tmp/72_dpi.pnm "${HOSTNAME}:10000/gocr_72_dpi"
+	curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary @tmp/72_dpi.png "${HOSTNAME}:10000/gocr_72_dpi"
 	rm -rf tmp
 
 client-108dpi:
 	mkdir -p tmp
 	shuf -n10 /usr/share/dict/american-english > tmp/10_words.txt
 	pango-view --dpi=108 --font=mono -qo tmp/108_dpi.png tmp/10_words.txt
-	pngtopnm tmp/108_dpi.png > tmp/108_dpi.pnm
-	curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary @tmp/108_dpi.pnm "${HOSTNAME}:10000/gocr_108_dpi"
+	curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary @tmp/108_dpi.png "${HOSTNAME}:10000/gocr_108_dpi"
 	rm -rf tmp
 
 client-144dpi:
 	mkdir -p tmp
 	shuf -n10 /usr/share/dict/american-english > tmp/10_words.txt
 	pango-view --dpi=144 --font=mono -qo tmp/144_dpi.png tmp/10_words.txt
-	pngtopnm tmp/144_dpi.png > tmp/144_dpi.pnm
-	curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary @tmp/144_dpi.pnm "${HOSTNAME}:10000/gocr_144_dpi"
+	curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary @tmp/144_dpi.png "${HOSTNAME}:10000/gocr_144_dpi"
 	rm -rf tmp

tests/gocr/by_dpi/README.md

@@ -1,6 +1,6 @@
 # OCR varied by DPI
 
-Generates pnm image files containing random words from the wamerican dictionary.
+Generates png image files containing random words from the wamerican dictionary.
 
 The number of words are held constant at 100, and the `mono` font is used for all runs. The dpi of the image is varied to judge how more visual details affects OCR accuracy and performance.
 

tests/gocr/by_dpi/run.sh

@@ -14,23 +14,12 @@ source percentiles_table.sh || exit 1
 source validate_dependencies.sh || exit 1
 
 experiment_client() {
-	local -ri iteration_count=100
-	local -ri word_count=100
-
-	if (($# != 2)); then
-		panic "invalid number of arguments \"$1\""
-		return 1
-	elif [[ -z "$1" ]]; then
-		panic "hostname \"$1\" was empty"
-		return 1
-	elif [[ ! -d "$2" ]]; then
-		panic "directory \"$2\" does not exist"
-		return 1
-	fi
-
 	local -r hostname="$1"
 	local -r results_directory="$2"
 
+	local -ri iteration_count=100
+	local -ri word_count=100
+
 	# Perform Experiments
 	printf "Running Experiments\n"
 	local -ar dpis=(72 108 144)
@@ -39,17 +28,17 @@ experiment_client() {
 		[108]=/gocr_108_dpi
 		[144]=/gocr_144_dpi
 	)
+	
 	local words
 	for ((i = 0; i < iteration_count; i++)); do
 		words="$(shuf -n"$word_count" /usr/share/dict/american-english)"
 
 		for dpi in "${dpis[@]}"; do
 			pango-view --dpi="$dpi" --font=mono -qo "${dpi}"_dpi.png -t "$words"
-			pngtopnm "${dpi}"_dpi.png > "${dpi}"_dpi.pnm
 
-			result=$(curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary @"${dpi}"_dpi.pnm "$hostname:10000${dpi_to_path[$dpi]}" --silent -w "%{stderr}%{time_total}\n" 2>> "$results_directory/${dpi}_time.txt")
+			result=$(curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary @"${dpi}"_dpi.png "$hostname:10000${dpi_to_path[$dpi]}" --silent -w "%{stderr}%{time_total}\n" 2>> "$results_directory/${dpi}_time.txt")
 
-			rm "${dpi}"_dpi.png "${dpi}"_dpi.pnm
+			rm "${dpi}"_dpi.png
 
 			# Logs the number of words that don't match
 			echo "word count: $word_count" >> "$results_directory/${dpi}_full_results.txt"
@@ -94,6 +83,6 @@ experiment_client() {
 }
 
 # Validate that required tools are in path
-validate_dependencies curl shuf pango-view pngtopnm diff
+validate_dependencies curl shuf pango-view diff
 
 framework_init "$@"

tests/gocr/by_dpi/spec.json

@@ -6,21 +6,18 @@
 			{
 				"route": "/gocr_72_dpi",
 				"path": "gocr.wasm.so",
-				"expected-execution-us": 5000,
 				"relative-deadline-us": 36000,
 				"http-resp-content-type": "text/plain"
 			},
 			{
 				"route": "/gocr_108_dpi",
 				"path": "gocr.wasm.so",
-				"expected-execution-us": 5000,
 				"relative-deadline-us": 36000,
 				"http-resp-content-type": "text/plain"
 			},
 			{
 				"route": "/gocr_144_dpi",
 				"path": "gocr.wasm.so",
-				"expected-execution-us": 5000,
 				"relative-deadline-us": 36000,
 				"http-resp-content-type": "text/plain"
 			}

tests/gocr/by_font/.gitignore

@@ -1,4 +1,3 @@
-*.pnm
 *.png
 *.csv
 *.txt

tests/gocr/by_font/Makefile

@@ -16,22 +16,19 @@ client-mono:
 	mkdir -p tmp
 	shuf -n10 /usr/share/dict/american-english > tmp/10_words.txt
 	pango-view --font="DejaVu Sans Mono" -qo tmp/10_words.png tmp/10_words.txt
-	pngtopnm tmp/10_words.png > tmp/10_words.pnm
-	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@tmp/10_words.pnm" "${HOSTNAME}:10000/gocr_mono"
+	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@tmp/10_words.png" "${HOSTNAME}:10000/gocr_mono"
 	rm -rf tmp
 
 client-roboto:
 	mkdir -p tmp
 	shuf -n10 /usr/share/dict/american-english > tmp/10_words.txt
 	pango-view --font="Roboto" -qo tmp/10_words.png tmp/10_words.txt
-	pngtopnm tmp/10_words.png > tmp/10_words.pnm
-	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@tmp/10_words.pnm" "${HOSTNAME}:10000/gocr_roboto"
+	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@tmp/10_words.png" "${HOSTNAME}:10000/gocr_roboto"
 	rm -rf tmp
 
 client-cascadia:
 	mkdir -p tmp
 	shuf -n10 /usr/share/dict/american-english > tmp/10_words.txt
 	pango-view --font="Cascadia Code" -qo tmp/10_words.png tmp/10_words.txt
-	pngtopnm tmp/10_words.png > tmp/10_words.pnm
-	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@tmp/10_words.pnm" "${HOSTNAME}:10000/gocr_cascadia"
+	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@tmp/10_words.png" "${HOSTNAME}:10000/gocr_cascadia"
 	rm -rf tmp

tests/gocr/by_font/README.md

@@ -1,6 +1,6 @@
 # OCR varied by font
 
-Generates pnm image files containing random words from the wamerican dictionary.
+Generates png image files containing random words from the wamerican dictionary.
 
 The number of words is held constant at 100, but a different font is used to render the text. Generally, `mono` performs the best, followed by `URW Gothic`, followed by `Lobster Two`, which is a stylistic font that performs quite poorly. 
 

tests/gocr/by_font/run.sh

@@ -14,23 +14,12 @@ source percentiles_table.sh || exit 1
 source validate_dependencies.sh || exit 1
 
 experiment_client() {
-	local -ri iteration_count=100
-	local -ri word_count=100
-
-	if (($# != 2)); then
-		panic "invalid number of arguments \"$1\""
-		return 1
-	elif [[ -z "$1" ]]; then
-		panic "hostname \"$1\" was empty"
-		return 1
-	elif [[ ! -d "$2" ]]; then
-		panic "directory \"$2\" does not exist"
-		return 1
-	fi
-
 	local -r hostname="$1"
 	local -r results_directory="$2"
 
+	local -ri iteration_count=100
+	local -ri word_count=100
+
 	# Perform Experiments
 	printf "Running Experiments\n"
 	local -ra fonts=("DejaVu Sans Mono" "Roboto" "Cascadia Code")
@@ -39,6 +28,7 @@ experiment_client() {
 		["Roboto"]=/gocr_roboto
 		["Cascadia Code"]=/gocr_cascadia
 	)
+	
 	local words
 	for ((i = 1; i <= iteration_count; i++)); do
 		words="$(shuf -n"$word_count" /usr/share/dict/american-english)"
@@ -47,11 +37,10 @@ experiment_client() {
 			font_file="${font// /_}"
 
 			pango-view --font="$font" -qo "${font_file}_words.png" -t "$words" || exit 1
-			pngtopnm "${font_file}_words.png" > "${font_file}_words.pnm"
 
-			result=$(curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary @"${font_file}_words.pnm" "$hostname:10000${font_to_path[$font]}" --silent -w "%{stderr}%{time_total}\n" 2>> "$results_directory/${font_file}_time.txt")
+			result=$(curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary @"${font_file}_words.png" "$hostname:10000${font_to_path[$font]}" --silent -w "%{stderr}%{time_total}\n" 2>> "$results_directory/${font_file}_time.txt")
 
-			rm "${font_file}"_words.png "${font_file}"_words.pnm
+			rm "${font_file}"_words.png
 
 			# Logs the number of words that don't match
 			echo "font: $font_file" >> "$results_directory/${font_file}_full_results.txt"
@@ -95,6 +84,6 @@ experiment_client() {
 	csv_to_dat "$results_directory/success.csv" "$results_directory/latency.csv"
 }
 
-validate_dependencies curl shuf pango-view pngtopnm diff
+validate_dependencies curl shuf pango-view diff
 
 framework_init "$@"

tests/gocr/by_font/spec.json

@@ -6,21 +6,18 @@
 			{
 				"route": "/gocr_mono",
 				"path": "gocr.wasm.so",
-				"expected-execution-us": 5000,
 				"relative-deadline-us": 36000,
 				"http-resp-content-type": "text/plain"
 			},
 			{
 				"route": "/gocr_roboto",
 				"path": "gocr.wasm.so",
-				"expected-execution-us": 5000,
 				"relative-deadline-us": 36000,
 				"http-resp-content-type": "text/plain"
 			},
 			{
 				"route": "/gocr_cascadia",
 				"path": "gocr.wasm.so",
-				"expected-execution-us": 5000,
 				"relative-deadline-us": 36000,
 				"http-resp-content-type": "text/plain"
 			}

tests/gocr/by_word/Makefile

@@ -16,14 +16,12 @@ client-10:
 	mkdir -p tmp
 	shuf -n10 /usr/share/dict/american-english > tmp/10_words.txt
 	pango-view --font=mono -qo tmp/10_words.png tmp/10_words.txt
-	pngtopnm tmp/10_words.png > tmp/10_words.pnm
-	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@tmp/10_words.pnm" "${HOSTNAME}:10000/gocr_10_words"
+	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@tmp/10_words.png" "${HOSTNAME}:10000/gocr_10_words"
 	rm -rf tmp
 
 client-100:
 	mkdir -p tmp
 	shuf -n100 /usr/share/dict/american-english > tmp/100_words.txt
 	pango-view --font=mono -qo tmp/100_words.png tmp/100_words.txt
-	pngtopnm tmp/100_words.png > tmp/100_words.pnm
-	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@tmp/100_words.pnm" "${HOSTNAME}:10000/gocr_100_words"
+	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@tmp/100_words.png" "${HOSTNAME}:10000/gocr_100_words"
 	rm -rf tmp

tests/gocr/by_word/README.md

@@ -1,6 +1,6 @@
 # OCR varied by word
 
-Generates pnm image files containing random words from the wamerican dictionary.
+Generates png image files containing random words from the wamerican dictionary.
 
 The number of words are varied between 1, 10, 100 in order to assess how the size of text OCRed affects the execution profile of the sandbox serving the request.
 

tests/gocr/by_word/run.sh

@@ -7,7 +7,6 @@ export PATH="$__run_sh__bash_libraries_absolute_path:$PATH"
 
 source csv_to_dat.sh || exit 1
 source framework.sh || exit 1
-# source generate_gnuplots.sh || exit 1
 source get_result_count.sh || exit 1
 source panic.sh || exit 1
 source path_join.sh || exit 1
@@ -15,24 +14,11 @@ source percentiles_table.sh || exit 1
 source validate_dependencies.sh || exit 1
 
 experiment_client() {
-	local -ir iteration_count=100
-	local -ra word_counts=(1 10 100)
-
-	if (($# != 2)); then
-		panic "invalid number of arguments \"$1\""
-		return 1
-	elif [[ -z "$1" ]]; then
-		panic "hostname \"$1\" was empty"
-		return 1
-	elif [[ ! -d "$2" ]]; then
-		panic "directory \"$2\" does not exist"
-		return 1
-	fi
-
 	local -r hostname="$1"
 	local -r results_directory="$2"
 
-	# Write Headers to CSV files
+	local -ir iteration_count=100
+	local -ra word_counts=(1 10 100)
 
 	local -Ar word_count_to_path=(
 		["1_words"]=/gocr_1_word
@@ -48,14 +34,13 @@ experiment_client() {
 			word_count_file="${word_count}_words"
 
 			pango-view --font=mono -qo "$word_count_file.png" -t "$words" || exit 1
-			pngtopnm "$word_count_file.png" > "$word_count_file.pnm" || exit 1
 
-			result=$(curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary @"$word_count_file.pnm" "$hostname:10000${word_count_to_path[$word_count_file]}" --silent -w "%{stderr}%{time_total}\n" 2>> "$results_directory/${word_count_file}_time.txt")
+			result=$(curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary @"$word_count_file.png" "$hostname:10000${word_count_to_path[$word_count_file]}" --silent -w "%{stderr}%{time_total}\n" 2>> "$results_directory/${word_count_file}_time.txt")
 
 			# If the OCR does not produce a guess, fail
 			[[ -z "$result" ]] && exit 1
 
-			rm "$word_count_file.png" "$word_count_file.pnm"
+			rm "$word_count_file.png"
 
 			# Logs the number of words that don't match
 			# Also tees the full diff into a separate file
@@ -68,6 +53,7 @@ experiment_client() {
 	done
 
 	# Process Results
+	# Write Headers to CSV files
 	printf "words,Success_Rate\n" >> "$results_directory/success.csv"
 	percentiles_table_header "$results_directory/latency.csv" "words"
 
@@ -100,6 +86,6 @@ experiment_client() {
 	csv_to_dat "$results_directory/success.csv" "$results_directory/latency.csv"
 }
 
-validate_dependencies curl shuf pango-view pngtopnm diff
+validate_dependencies curl shuf pango-view diff
 
 framework_init "$@"

tests/gocr/by_word/spec.json

@@ -6,21 +6,18 @@
 			{
 				"route": "/gocr_1_word",
 				"path": "gocr.wasm.so",
-				"expected-execution-us": 5000,
 				"relative-deadline-us": 36000,
 				"http-resp-content-type": "text/plain"
 			},
 			{
 				"route": "/gocr_10_words",
 				"path": "gocr.wasm.so",
-				"expected-execution-us": 5000,
 				"relative-deadline-us": 36000,
 				"http-resp-content-type": "text/plain"
 			},
 			{
 				"route": "/gocr_100_words",
 				"path": "gocr.wasm.so",
-				"expected-execution-us": 5000,
 				"relative-deadline-us": 36000,
 				"http-resp-content-type": "text/plain"
 			}

tests/gocr/fivebyeight/Makefile

@@ -13,4 +13,4 @@ debug:
 	SLEDGE_DISABLE_PREEMPTION=true SLEDGE_NWORKERS=1 LD_LIBRARY_PATH=${SLEDGE_BINARY_DIR} gdb ${SLEDGE_BINARY_DIR}/sledgert --eval-command="run spec.json"
 
 client:
-	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@5x8.pnm" "${HOSTNAME}:10000/gocr"
+	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@5x8.png" "${HOSTNAME}:10000/gocr"

tests/gocr/fivebyeight/README.md

@@ -1,2 +0,0 @@
-pnm file was generated as follows
-`pngtopnm 5x8.png >5x8.pnm`

tests/gocr/fivebyeight/run.sh

@@ -23,7 +23,7 @@ experiment_client() {
 
 	local result
 	for ((i = 0; i < total_count; i++)); do
-		result=$(curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@5x8.pnm" "$hostname:10000/gocr" 2> /dev/null)
+		result=$(curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@5x8.png" "$hostname:10000/gocr" 2> /dev/null)
 		if [[ "$result" == "$expected_result" ]]; then
 			((success_count++))
 		else

tests/gocr/fivebyeight/spec.json

@@ -6,7 +6,6 @@
 			{
 				"route": "/gocr",
 				"path": "gocr.wasm.so",
-				"expected-execution-us": 5000,
 				"relative-deadline-us": 36000,
 				"http-resp-content-type": "text/plain"
 			}

tests/gocr/handwriting/Makefile

@@ -13,4 +13,4 @@ debug:
 	LD_LIBRARY_PATH=${SLEDGE_BINARY_DIR} gdb ${SLEDGE_BINARY_DIR}/sledgert --eval-command="run spec.json"
 
 client:
-	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@handwrt1.pnm" "${HOSTNAME}:10000/gocr"
+	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@handwrt.png" "${HOSTNAME}:10000/gocr"

tests/gocr/handwriting/run.sh

@@ -22,7 +22,7 @@ experiment_client() {
 	local -ir total_count=10
 
 	for ((i = 0; i < total_count; i++)); do
-		result=$(curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@handwrt1.pnm" "$hostname:10000/gocr" 2> /dev/null)
+		result=$(curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@handwrt.png" "$hostname:10000/gocr" 2> /dev/null)
 		if [[ "$result" == "$expected_result" ]]; then
 			((success_count++))
 		else

tests/gocr/handwriting/spec.json

@@ -6,7 +6,6 @@
 			{
 				"route": "/gocr",
 				"path": "gocr.wasm.so",
-				"expected-execution-us": 5000,
 				"relative-deadline-us": 36000,
 				"http-resp-content-type": "text/plain"
 			}

tests/gocr/hyde/Makefile

@@ -16,10 +16,4 @@ valgrind:
 	SLEDGE_DISABLE_PREEMPTION=true SLEDGE_NWORKERS=1 LD_LIBRARY_PATH=${SLEDGE_BINARY_DIR} valgrind --leak-check=full --max-stackframe=11150456 --run-libc-freeres=no --run-cxx-freeres=no ${SLEDGE_BINARY_DIR}/sledgert spec.json
 
 client:
-	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@hyde.pnm" "${HOSTNAME}:10000/gocr"
-
-client2:
-	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@../handwriting/handwrt1.pnm" "${HOSTNAME}:10000/gocr"
-
-client3:
-	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@../fivebyeight/5x8.pnm" "${HOSTNAME}:10000/gocr"
+	curl  -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@hyde.png" "${HOSTNAME}:10000/gocr"

tests/gocr/hyde/README.md

@@ -1,5 +1,3 @@
 # A Page from Dr. Jekyll and Mister Hyde
 
-This is a larger OCR example. The \*.pnm file is 5mb or so.
-
-In the future, this OCR example might be a good use case for a "step-wise" function, as the file is substantially smaller before conversion to the \*.pnm format.
+This is a larger OCR example.

tests/gocr/hyde/run.sh

@@ -21,7 +21,7 @@ experiment_client() {
 	local -ir total_count=10
 
 	for ((i = 0; i < total_count; i++)); do
-		result=$(curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@hyde.pnm" "$hostname:10000/gocr" 2> /dev/null)
+		result=$(curl -H 'Expect:' -H "Content-Type: text/plain" --data-binary "@hyde.png" "$hostname:10000/gocr" 2> /dev/null)
 		if [[ "$result" == "$expected_result" ]]; then
 			((success_count++))
 		else

tests/gocr/hyde/spec.json

@@ -6,7 +6,6 @@
 			{
 				"route": "/gocr",
 				"path": "gocr.wasm.so",
-				"expected-execution-us": 5000,
 				"relative-deadline-us": 360000,
 				"http-resp-content-type": "text/plain"
 			}