Skip to contentSkip to conversation

support/scripts/pkg-stats: URL checking support

PATCH New Matt Weber 4 participants 8 messages Oct 2, 2018 – Oct 9, 2018

lore
Share & tools
lore

Matt Weber

author
#0 lore raw

PATCH v3 1/2 support/scripts/pkg-stats: URL checking support

- Adds support to check if a package has a URL and if that URL is valid by doing a header request.
- Reports this information as part of the generated html output

The URL data is currently gathered from the URL string provided in the Kconfig help sections for each package.

This check helps ensure the URLs are valid and can be used for other scripting purposes as the product's home site/URL. CPE XML generation is an example of a case that could use this product URL as part of an automated update generation script.

CC Ricardo Martincoski <[email protected]>

---
Changes
v2 -> v3
[Ricardo
- Fixed flake8 warnings of extra lines and unused variable
- Adjusted regex to be compiled once and used example to simplify finding the url and cleaning up the string
- Removed checking for help section and instead used the suggested improved regex to find the URL.

v1 -> v2
- Dropped disabling of SSL cert verifing

[Thomas
- Moved URL report to new column
- Added color coding
- Better info on if a package doesn't have a Config.in to search or the Config.in is missing the URL

---
 support/scripts/pkg-stats | 57 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 57 insertions(+)
support/scripts/pkg-stats +57 -0
index b7b00e8..c841e09 100755
@@ -24,8 +24,10 @@ from collections import defaultdict
 import re
 import subprocess
 import sys
+import requests  # URL checking
 
 INFRA_RE = re.compile("\$\(eval \$\(([a-z-]*)-package\)\)")
+URL_RE = re.compile("\s*https?://\S*\s*$")
 
 
 class Package:
@@ -43,10 +45,29 @@ class Package:
         self.patch_count = 0
         self.warnings = 0
         self.current_version = None
+        self.url = None
+        self.url_status = None
 
     def pkgvar(self):
         return self.name.upper().replace("-", "_")
 
+    def set_url(self):
+        """
+        Fills in the .url field
+        """
+        self.url_status = "No Config.in"
+        for filename in os.listdir(os.path.dirname(self.path)):
+            if fnmatch.fnmatch(filename, 'Config.*'):
+                fp = open(os.path.join(os.path.dirname(self.path), filename), "r")
+                for config_line in fp:
+                    if URL_RE.match(config_line):
+                        self.url = config_line.strip()
+                        self.url_status = "Found"
+                        fp.close()
+                        return
+                self.url_status = "Missing"
+                fp.close()
+
     def set_infra(self):
         """
         Fills in the .infras field
@@ -255,6 +276,16 @@ def package_init_make_info():
         Package.all_versions[pkgvar] = value
 
 
+def check_url_status(pkg):
+    if pkg.url_status != "Missing" and pkg.url_status != "No Config.in":
+        try:
+            url_status_code = requests.head(pkg.url, timeout=5).status_code
+            if url_status_code >= 400:
+                pkg.url_status = "Invalid(%s)" % str(url_status_code)
+        except requests.exceptions.RequestException:
+            return
+
+
 def calculate_stats(packages):
     stats = defaultdict(int)
     for pkg in packages:
@@ -311,6 +342,15 @@ td.somepatches {
 td.lotsofpatches {
   background: #ff9a69;
 }
+td.good_url {
+  background: #d2ffc4;
+}
+td.missing_url {
+  background: #ffd870;
+}
+td.invalid_url {
+  background: #ff9a69;
+}
 </style>
 <title>Statistics of Buildroot packages</title>
 </head>
@@ -422,6 +462,20 @@ def dump_html_pkg(f, pkg):
     f.write("  <td class=\"%s\">%d</td>\n" %
             (" ".join(td_class), pkg.warnings))
 
+    # URL status
+    td_class = ["centered"]
+    url_str = pkg.url_status
+    if pkg.url_status == "Missing" or pkg.url_status == "No Config.in":
+        td_class.append("missing_url")
+    elif pkg.url_status.startswith("Invalid"):
+        td_class.append("invalid_url")
+        url_str = "<a href=%s>%s</a>" % (pkg.url, pkg.url_status)
+    else:
+        td_class.append("good_url")
+        url_str = "<a href=%s>Link</a>" % pkg.url
+    f.write("  <td class=\"%s\">%s</td>\n" %
+            (" ".join(td_class), url_str))
+
     f.write(" </tr>\n")
 
 
@@ -437,6 +491,7 @@ def dump_html_all_pkgs(f, packages):
 <td class=\"centered\">Hash file</td>
 <td class=\"centered\">Current version</td>
 <td class=\"centered\">Warnings</td>
+<td class=\"centered\">URL</td>
 </tr>
 """)
     for pkg in sorted(packages):
@@ -517,6 +572,8 @@ def __main__():
         pkg.set_patch_count()
         pkg.set_check_package_warnings()
         pkg.set_current_version()
+        pkg.set_url()
+        check_url_status(pkg)
     print("Calculate stats")
     stats = calculate_stats(packages)
     print("Write HTML")
Signature · 1 line
1.9.1
author
#1 lore raw

PATCH v3 2/2 support/scripts/pkg-stats: URL check using threads

Adds a pool of worker threads to accelerate connection testing.

~7.5MB and 2% CPU per thread on a Intel i5-3230M CPU @ 2.60GHz.

Runtime is ~3min in parallel vs ~15min.

CC Ricardo Martincoski <[email protected]>

---
Changes
v2 -> v3
[Ricardo
- Updated the timeout for header request testing to 30sec as there maybe corner cases with a smaller value when running requests in parallel on a busy machine

v1 -> v2
- New patch

---
 support/scripts/pkg-stats | 25 +++++++++++++++++++------
 1 file changed, 19 insertions(+), 6 deletions(-)
support/scripts/pkg-stats +19 -6
index c841e09..b615001 100755
@@ -25,6 +25,7 @@ import re
 import subprocess
 import sys
 import requests  # URL checking
+from multiprocessing import Pool
 
 INFRA_RE = re.compile("\$\(eval \$\(([a-z-]*)-package\)\)")
 URL_RE = re.compile("\s*https?://\S*\s*$")
@@ -47,6 +48,7 @@ class Package:
         self.current_version = None
         self.url = None
         self.url_status = None
+        self.url_worker = None
 
     def pkgvar(self):
         return self.name.upper().replace("-", "_")
@@ -276,14 +278,24 @@ def package_init_make_info():
         Package.all_versions[pkgvar] = value
 
 
-def check_url_status(pkg):
-    if pkg.url_status != "Missing" and pkg.url_status != "No Config.in":
+def check_url_status_worker(url, url_status):
+    if url_status != "Missing" and url_status != "No Config.in":
         try:
-            url_status_code = requests.head(pkg.url, timeout=5).status_code
+            url_status_code = requests.head(url, timeout=30).status_code
             if url_status_code >= 400:
-                pkg.url_status = "Invalid(%s)" % str(url_status_code)
+                return "Invalid(%s)" % str(url_status_code)
         except requests.exceptions.RequestException:
-            return
+            return "Invalid(Err)"
+        return "Ok"
+    return url_status
+
+
+def check_package_urls(packages):
+    Package.pool = Pool(processes=64)
+    for pkg in packages:
+        pkg.url_worker = pkg.pool.apply_async(check_url_status_worker, (pkg.url, pkg.url_status))
+    for pkg in packages:
+        pkg.url_status = pkg.url_worker.get(timeout=3600)
 
 
 def calculate_stats(packages):
@@ -573,7 +585,8 @@ def __main__():
         pkg.set_check_package_warnings()
         pkg.set_current_version()
         pkg.set_url()
-        check_url_status(pkg)
+    print("Checking URL status")
+    check_package_urls(packages)
     print("Calculate stats")
     stats = calculate_stats(packages)
     print("Write HTML")
Signature · 1 line
1.9.1
#3 lore raw

PATCH v3 2/2 support/scripts/pkg-stats: URL check using threads

Hello Ricardo,

Quoted 4 lines · Ricardo Martincoski
Quoted 2 lines · Matt Weber

Adds a pool of worker threads to accelerate connection testing.

Quoted 8 lines
Quoted 6 lines

~7.5MB and 2% CPU per thread on a Intel i5-3230M CPU @ 2.60GHz.

Runtime is ~3min in parallel vs ~15min.

CC Ricardo Martincoski <[email protected]>

So this threading works for you? I think it is modeled after what I initially did to retrieve the upstream version of the packages, and you said it wasn't working well for you.

Best regards,

Thomas

Signature · 3 lines
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
#4 lore raw

PATCH v3 2/2 support/scripts/pkg-stats: URL check using threads

Hello Thomas,

Quoted 1 line · Thomas Petazzoni

On Tue, 02 Oct 2018 21:50:04 -0300, Ricardo Martincoski wrote:

[snip]

Quoted 5 lines
Quoted 1 line

So this threading works for you? I think it is modeled after what I initially did to retrieve the upstream version of the packages, and you said it wasn't working well for you.

Yes. It is working for me.

Actually you used threading + Queue, and I suggested you to switch to use multiprocessing.
https://patchwork.ozlabs.org/patch/890236/

Matt uses multiprocessing in this patch.

Regards,
Ricardo

#5 lore raw

PATCH v3 2/2 support/scripts/pkg-stats: URL check using threads

In reply to Matt Weber

Hello,

Quoted 9 lines · Matt Weber

Adds a pool of worker threads to accelerate connection testing.

~7.5MB and 2% CPU per thread on a Intel i5-3230M CPU @ 2.60GHz.

Runtime is ~3min in parallel vs ~15min.

CC Ricardo Martincoski <[email protected]>
Quoted 7 lines

---
Changes
v2 -> v3
[Ricardo
- Updated the timeout for header request testing to 30sec as there maybe corner cases with a smaller value when running requests in parallel on a busy machine

Applied to master, thanks.

Thomas

Signature · 3 lines
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
#6 lore raw

PATCH v3 1/2 support/scripts/pkg-stats: URL checking support

In reply to Matt Weber

Hello,

Quoted 7 lines · Matt Weber

- Adds support to check if a package has a URL and if that URL is valid by doing a header request.
- Reports this information as part of the generated html output

The URL data is currently gathered from the URL string provided in the Kconfig help sections for each package.

Quoted 7 lines

This check helps ensure the URLs are valid and can be used for other scripting purposes as the product's home site/URL. CPE XML generation is an example of a case that could use this product URL as part of an automated update generation script.

CC Ricardo Martincoski <[email protected]>

Regards,
Ricardo

#7 lore raw

PATCH v3 1/2 support/scripts/pkg-stats: URL checking support

In reply to Matt Weber

Hello,

Quoted 14 lines · Matt Weber

- Adds support to check if a package has a URL and if that URL is valid by doing a header request.
- Reports this information as part of the generated html output

The URL data is currently gathered from the URL string provided in the Kconfig help sections for each package.

This check helps ensure the URLs are valid and can be used for other scripting purposes as the product's home site/URL. CPE XML generation is an example of a case that could use this product URL as part of an automated update generation script.

CC Ricardo Martincoski <[email protected]>
Quoted 9 lines

---
Changes
v2 -> v3
[Ricardo
- Fixed flake8 warnings of extra lines and unused variable
- Adjusted regex to be compiled once and used example to simplify finding the url and cleaning up the string
- Removed checking for help section and instead used the suggested improved regex to find the URL.

Applied to master, thanks. I have just changed the header of the column from "URL" to "Upstream URL".

Thomas

Signature · 3 lines
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com