create_perf_json: Migrate to pathlib.Path for file operations

[edwarddavidbaker]

Previously event files were hosted at https://download.01.org/perfmon and fetching them with urllib was necessary. After migrating to GitHub we no longer need to fetch files using urllib.

• Script arguments
  • Removed base path argument. At this point in time, create_perf_json.py only operates on files in this repository.
  • Add Path type to argument parser.
• Migrate to Path
  • Use pathlib.Path to construct file and folder paths.
  • Updated type annotations with Path.
  • Replaced urlopen() with open().
  • Removed unused imports os and urllib.
  • Switched a few try: blocks to if <path>.is_file().
• CI
  • Execute metric tests.

[edwarddavidbaker]

Thanks Ian. Adding a few notes comparing the console and output directory contents. Output is the same. Console output has slight variations due to paths instead of file://.

# Generate output from main branch.
ed@ebaker-MOBL3:~/perfmon/public/perfmon/scripts$ git checkout main
ed@ebaker-MOBL3:~/perfmon/public/perfmon/scripts$ mkdir main
ed@ebaker-MOBL3:~/perfmon/public/perfmon/scripts$ python create_perf_json.py -vvv --outdir ./main/perf > ./main/output.txt

# Generate output from pull request.
ed@ebaker-MOBL3:~/perfmon/public/perfmon/scripts$ git checkout create_perf_json_pathlib 
ed@ebaker-MOBL3:~/perfmon/public/perfmon/scripts$ mkdir pr
ed@ebaker-MOBL3:~/perfmon/public/perfmon/scripts$ python create_perf_json.py -vvv --outdir ./pr/perf > ./pr/output.txt

Output content is the same.

ed@ebaker-MOBL3:~/perfmon/public/perfmon/scripts$ diff -rup main/perf/ pr/perf/
ed@ebaker-MOBL3:~/perfmon/public/perfmon/scripts$

Verbose logging is slightly different, but essentially the same.

ed@ebaker-MOBL3:~/perfmon/public/perfmon/scripts$ diff -rup main/output.txt pr/output.txt
<snip>
 Parsed models:
 ADL / alderlake
 	models=['GenuineIntel-6-97', 'GenuineIntel-6-9A', 'GenuineIntel-6-B7', 'GenuineIntel-6-BA', 'GenuineIntel-6-BF']
 	files:
-		atom = file:///home/ed/Documents/perfmon/public/perfmon/scripts/../ADL/events/alderlake_gracemont_core.json
-		core = file:///home/ed/Documents/perfmon/public/perfmon/scripts/../ADL/events/alderlake_goldencove_core.json
-		uncore = file:///home/ed/Documents/perfmon/public/perfmon/scripts/../ADL/events/alderlake_uncore.json
-		uncore experimental = file:///home/ed/Documents/perfmon/public/perfmon/scripts/../ADL/events/alderlake_uncore_experimental.json
-		tma metrics = file:///home/ed/Documents/perfmon/public/perfmon/scripts/../TMA_Metrics-full.csv
-		e-core tma metrics = file:///home/ed/Documents/perfmon/public/perfmon/scripts/../E-core_TMA_Metrics.csvADLN / alderlaken
+		atom = /home/ed/Documents/perfmon/public/perfmon/ADL/events/alderlake_gracemont_core.json
+		core = /home/ed/Documents/perfmon/public/perfmon/ADL/events/alderlake_goldencove_core.json
+		uncore = /home/ed/Documents/perfmon/public/perfmon/ADL/events/alderlake_uncore.json
+		uncore experimental = /home/ed/Documents/perfmon/public/perfmon/ADL/events/alderlake_uncore_experimental.json
+		tma metrics = /home/ed/Documents/perfmon/public/perfmon/TMA_Metrics-full.csv
+		e-core tma metrics = /home/ed/Documents/perfmon/public/perfmon/E-core_TMA_Metrics.csvADLN / alderlaken
 	models=['GenuineIntel-6-BE']
 	files:
<snip>
@@ -13508,8 +13508,8 @@ Updated tma_port_4 from
 "UOPS_DISPATCHED_PORT.PORT_4 / tma_info_core_core_clks"
 to
 "tma_store_op_utilization"
-Creating event json for CLX in ./main/perf/cascadelakex
-Generating core events from file:///home/ed/Documents/perfmon/public/perfmon/scripts/../CLX/events/cascadelakex_core.json
+Creating event json for CLX in /home/ed/Documents/perfmon/public/perfmon/scripts/pr/perf/cascadelakex
+Generating core events from /home/ed/Documents/perfmon/public/perfmon/CLX/events/cascadelakex_core.json
 Read perfmon event:
 event_name: INST_RETIRED.ANY, sample_after_value: 2000003, umask: 0x1, brief_description: Instructions retired from execution., public_description: Counts the number of instructions retired from execution. For instructions that consist of multiple micro-ops, Counts the retirement of the last micro-op of the instruction. Counting continues during hardware interrupts, traps, and inside interrupt handlers. Notes: INST_RETIRED.ANY is counted by a designated fixed counter, leaving the four (eight when Hyperthreading is disabled) programmable counters available for other events. INST_RETIRED.ANY_P is counted by a programmable counter and it is an architectural performance event. Counting: Faulting executions of GETSEC/VM entry/VM Exit/MWait will not count as retired instructions., topic: Pipeline
 Read perfmon event:
<snip>