GuilloteauQ/study-docker-repro-longevity
1
0
Fork 0
Code Issues 6 Pull Requests Actions Projects Releases Wiki Activity

starting cleaning snakefile

Browse Source
This commit is contained in:
Quentin Guilloteau 2024-08-28 14:35:13 +02:00
parent d57f8b019e
commit 025a16b62c
1 changed files with 15 additions and 68 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

83
workflow/Snakefile
View File

@ -34,9 +34,10 @@ SOFTENV_ANALYSIS_DIRS = " ".join(expand(f"{PREFIX}/pkgs/{{artifact}}",
))
PLOT_DIR = config["plot_dir"]
PLOT_SCRIPT = "plot/plot.r"
PLOT_HEADERS = {
"softenv": "dpkg rpm pacman pip conda git misc",
#"softenv": "dpkg rpm pacman pip conda git misc",
"sources_stats": "dpkg rpm pacman pip conda git misc",
"pkgs_changes": "dpkg rpm pacman pip conda git misc",
"build_status": "success package_install_failed baseimage_unavailable artifact_unavailable dockerfile_not_found script_crash job_time_exceeded unknown_error",
"artifact": "available unavailable changed"
}
@ -157,78 +158,24 @@ rule analysis_aggregate:
date = DATE
)
rule pkgschgs_aggregate:
rule aggregate_by_type:
input:
f"{ANALYSIS_DIR}/pkgs_changes/{{date}}.csv"
data=f"{ANALYSIS_DIR}/{{type}}/{{date}}.csv",
script="workflow/scripts/aggregate_wrapper.sh"
output:
f"{ANALYSIS_DIR}/pkgs_changes/aggregated/{{date}}.csv"
f"{ANALYSIS_DIR}/{{type}}/aggregated/{{date}}.csv"
shell:
f"{AGGREGATE_WRAPPER} {ANALYSIS_DIR}/pkgs_changes {{output}}"
rule srcsstats_aggregate:
input:
f"{ANALYSIS_DIR}/sources_stats/{{date}}.csv"
output:
f"{ANALYSIS_DIR}/sources_stats/aggregated/{{date}}.csv"
shell:
f"{AGGREGATE_WRAPPER} {ANALYSIS_DIR}/sources_stats {{output}}"
rule artifact_aggregate:
input:
f"{ANALYSIS_DIR}/artifact/{{date}}.csv"
output:
f"{ANALYSIS_DIR}/artifact/aggregated/{{date}}.csv"
shell:
f"{AGGREGATE_WRAPPER} {ANALYSIS_DIR}/artifact {{output}}"
rule buildstatus_aggregate:
input:
f"{ANALYSIS_DIR}/build_status/{{date}}.csv"
output:
f"{ANALYSIS_DIR}/build_status/aggregated/{{date}}.csv"
shell:
f"{AGGREGATE_WRAPPER} {ANALYSIS_DIR}/build_status {{output}}"
f"{input.script} {ANALYSIS_DIR}/{{type}} {{output}}"
# Plot:
rule all_plot:
rule plot:
input:
expand(f"{ANALYSIS_DIR}/{{cat}}/aggregated/{{date}}.csv",
cat = ANALYSIS_CATS,
date = DATE
)
rule line_plot:
input:
sources_stats = f"{ANALYSIS_DIR}/sources_stats/aggregated/{{date}}.csv",
pkgs_changes = f"{ANALYSIS_DIR}/pkgs_changes/aggregated/{{date}}.csv",
build_status = f"{ANALYSIS_DIR}/build_status/aggregated/{{date}}.csv",
artifact = f"{ANALYSIS_DIR}/artifact/aggregated/{{date}}.csv"
script = "plot/plot.r",
data = f"{ANALYSIS_DIR}/{{type}}/aggregated/{{date}}.csv",
output:
sources_stats = f"{ANALYSIS_DIR}/sources_stats/plot/line/{{date}}.pdf",
pkgs_changes = f"{ANALYSIS_DIR}/pkgs_changes/plot/line/{{date}}.pdf",
build_status = f"{ANALYSIS_DIR}/build_status/plot/line/{{date}}.pdf",
artifact = f"{ANALYSIS_DIR}/artifact/plot/line/{{date}}.pdf"
f"{ANALYSIS_DIR}/{{type}}/{{plot}}/{{date}}.pdf"
params:
header = lambda w: PLOT_HEADERS[w.type]
shell:
f"""
Rscript {PLOT_SCRIPT} line {{input.sources_stats}} {{output.sources_stats}} {PLOT_HEADERS["softenv"]} timestamp
Rscript {PLOT_SCRIPT} line {{input.pkgs_changes}} {{output.pkgs_changes}} {PLOT_HEADERS["softenv"]} timestamp
Rscript {PLOT_SCRIPT} line {{input.build_status}} {{output.build_status}} {PLOT_HEADERS["build_status"]} timestamp
Rscript {PLOT_SCRIPT} line {{input.artifact}} {{output.artifact}} {PLOT_HEADERS["artifact"]} timestamp
"""
rule bar_plot:
input:
sources_stats = f"{ANALYSIS_DIR}/sources_stats/aggregated/{{date}}.csv",
build_status = f"{ANALYSIS_DIR}/build_status/aggregated/{{date}}.csv",
artifact = f"{ANALYSIS_DIR}/artifact/aggregated/{{date}}.csv"
output:
sources_stats = f"{ANALYSIS_DIR}/sources_stats/plot/bar/{{date}}.pdf",
build_status = f"{ANALYSIS_DIR}/build_status/plot/bar/{{date}}.pdf",
artifact = f"{ANALYSIS_DIR}/artifact/plot/bar/{{date}}.pdf"
shell:
f"""
Rscript {PLOT_SCRIPT} bar {{input.sources_stats}} {{output.sources_stats}} {PLOT_HEADERS["softenv"]} timestamp
Rscript {PLOT_SCRIPT} bar {{input.build_status}} {{output.build_status}} {PLOT_HEADERS["build_status"]} timestamp
Rscript {PLOT_SCRIPT} bar {{input.artifact}} {{output.artifact}} {PLOT_HEADERS["artifact"]} timestamp
"""
"Rscript {input.script} {wildcards.plot} {input.data} {output} {params.header} timestamp"