kthutils 1.40__tar.gz → 1.41__tar.gz

{kthutils-1.40 → kthutils-1.41}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: kthutils
-Version: 1.40
+Version: 1.41
 Summary: Various tools for automation at KTH
 License-Expression: MIT
 License-File: LICENSE

{kthutils-1.40 → kthutils-1.41}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "kthutils"
-version = "1.40"
+version = "1.41"
 description = "Various tools for automation at KTH"
 authors = [
   {name="Daniel Bosk", email="dbosk@kth.se"}

{kthutils-1.40 → kthutils-1.41}/src/kthutils/forms.nw

@@ -819,46 +819,90 @@ courses in the same way; only the columns the fields live in differ.
 We therefore lift the body of the format string into its own chunk,
 [[<<rewriter format string body>>]], so both rewriters share a single
 definition and cannot drift apart when we tweak the reporting commands.
+
+The body passes through two interpreters before it runs:
+first the shell executing this setup script, then Python's [[str.format]]
+when a row is rewritten.
+Each has its own escaping (backslashes and [[$]] for the shell, doubled
+braces for [[format]]), and the body is a shell script itself, full of
+[[$]], quotes and line continuations.
+Escaping all of that for the setup-time shell would make the chunk
+unreadable and easy to break.
+We therefore feed the body through a quoted heredoc: with the delimiter
+quoted, the shell copies the text verbatim into the configuration, so the
+chunk below is exactly what ends up in the config and only the [[format]]
+braces remain to think about.
 <<[[restlabbsetup.sh]]>>=
-kthutils config ${RESTLABB_REWRITER}.format_string -s "
+kthutils config ${RESTLABB_REWRITER}.format_string -s "$(cat <<'_EOF'
 <<rewriter format string body>>
-"
+_EOF
+)"
 @
 
-The body echoes the parsed fields for the log, then---unless the grader
+The body echoes the fields for the log, then---unless the grader
 already reported the result in Canvas---runs [[canvaslms grade]], and
 finally reports the aggregated result to Ladok.
+
+The fields that go through a regex (the course nick and year, the student,
+the grader, and the grade, assignment and assignment group extracted from
+the module column) can be mis-parsed when a TA writes something we did not
+anticipate.
+When that happens we fix the value by hand before running the commands, and
+we want to fix it in one place, not at every use.
+So each such field is assigned to a shell variable once and the commands
+only use the variable.
+Fields copied verbatim from a column, such as [[in_canvas]] (\enquote{Ja}
+or \enquote{Nej}) and [[passed]], cannot be mis-parsed, so they stay inline.
+
+We place each assignment directly after the echo of the raw column it was
+parsed from: the reader then sees the input and the parsed result next to
+each other and can check the parse on the spot.
+The module column feeds three parsed fields, so all three follow the
+[[Module:]] line.
+The student and grader columns are only used through their regexes, so
+there the assignment comes first and the echo prints the variable.
+Assigned values are single-quoted since a parsed value can contain shell
+metacharacters (an assignment reads [[Laboration .(1|2).]]), and every use
+is double-quoted for the same reason.
+Since all rows are printed into one stream, each block re-assigns every
+variable before use, so the blocks stay independent when pasted together.
 <<rewriter format string body>>=
 echo 'Time:         {time}'
 echo 'Course:       {course}'
+course='{course_nick}{year}'
 echo 'Module:       {module}'
-echo 'Student:      {student}'
+assignment='{assignment}'
+assignment_group='{assignment_group}'
+grade='{grade}'
+student='{student}'
+echo "Student:      $student"
 echo 'Passed:       {passed}'
 echo 'Comments:     {comments}'
-echo 'Grader:       {grader}'
+grader='{grader}'
+echo "Grader:       $grader"
 echo 'In Canvas:    {in_canvas}'
 if [ '{in_canvas}' = 'Ja' ]; then
   echo 'Already in Canvas.'
 elif [ '{passed}' = 'Godkänt' ]; then
-  canvaslms grade \\
-    -c '{course_nick}{year}' \\
-    -a '{assignment}' \\
-    -u '{student}' \\
-    -g '{grade}' \\
-    -m 'OK {grader}'
+  canvaslms grade \
+    -c "$course" \
+    -a "$assignment" \
+    -u "$student" \
+    -g "$grade" \
+    -m "OK $grader"
 fi
 echo 'Canvas:'
-canvaslms submissions \\
-  -c '{course_nick}{year}' \\
-  -a '{assignment}' \\
-  -u '{student}'
+canvaslms submissions \
+  -c "$course" \
+  -a "$assignment" \
+  -u "$student"
 echo
-canvaslms results \\
-  -c '{course_nick}{year}' \\
-  -A '{assignment_group}' \\
-  -u '{student}' \\
-  -S canvaslms.grades.conjunctavgsurvey \\
-| sed -E 's/ ?[HV]T[0-9]{{2}}( \(.*\))?//' \\
+canvaslms results \
+  -c "$course" \
+  -A "$assignment_group" \
+  -u "$student" \
+  -S canvaslms.grades.conjunctavgsurvey \
+| sed -E 's/ ?[HV]T[0-9]{{2}}( \(.*\))?//' \
 | ladok -v report -f
 echo
 @
@@ -1154,7 +1198,7 @@ Each June we also run a \emph{lab week}, recording results in a separate KTH
 Form that we always name [[labweek<YEAR>]] (e.g.\ [[labweek2026]]).
 It reports results for the same courses in the same way as Restlabb, so it
 reuses the same format string and the same regexes.
-Only two things differ, and both follow from the form's layout.
+Only three things differ, and all follow from the form's layout.
 
 First, the columns sit in different positions.
 The lab week form lets a pair of students redovisa together, so on top of the
@@ -1175,13 +1219,20 @@ regexes---one extracting the first address, one the second---is joined into a
 parenthesised alternation when both match (and left as a bare address when
 only the first does).
 
+Third, the lab week form has no \enquote{passed?} question:
+the TAs only record results that passed, so every row is a pass.
+The shared format string still tests [[{passed}]] against \enquote{Godkänt}
+before grading in Canvas, so we supply that as a fixed [[value]] rather than
+from a column (see the [[value]] substitution in \cref{RewriteRow}).
+
 We clear any previous configuration first, in case we re-run these lines after
 an update.
 <<[[restlabbsetup.sh]]>>=
 kthutils config ${LABWEEK_REWRITER} -s ""
-kthutils config ${LABWEEK_REWRITER}.format_string -s "
+kthutils config ${LABWEEK_REWRITER}.format_string -s "$(cat <<'_EOF'
 <<rewriter format string body>>
-"
+_EOF
+)"
 @
 
 The verbatim columns and the shared regexes only differ from Restlabb in their
@@ -1204,6 +1255,7 @@ kthutils config ${LABWEEK_REWRITER}.substitutions.year.column -s 8
 kthutils config ${LABWEEK_REWRITER}.substitutions.year.regex \
   -s "<<year regex>>"
 kthutils config ${LABWEEK_REWRITER}.substitutions.in_canvas.column -s 12
+kthutils config ${LABWEEK_REWRITER}.substitutions.passed.value -s Godkänt
 kthutils config ${LABWEEK_REWRITER}.substitutions.grade.column -s 10
 kthutils config ${LABWEEK_REWRITER}.substitutions.grade.regex \
   <<grade regexes>>
@@ -1279,6 +1331,13 @@ else
 fi
 @
 
+When we extract the restlabb results, we must be logged into KTH.
+We can trigger that log in with [[ladok]].
+<<[[restlabb.sh]]>>=
+ME=de709f81-a867-11e7-8dbf-78e86dc2470c
+until ladok student $ME 2>&1 >/dev/null; do true; done
+@ This way we will not proceed until the MFA login has succeeded.
+
 Next we extract the new values since the last run (using the [[next]] command) 
 and [[grep]] for the relevant lines:
 namely those mentioning me or any of my courses.
@@ -1456,6 +1515,7 @@ delimiter: Annotated[str, delimiter_arg] = "\t",
 @
 
 \subsubsection{Rewriting a row}
+\label{RewriteRow}
 
 Now we want to do the actual rewriting.
 <<API for KTH Forms>>=
@@ -1539,6 +1599,18 @@ def test_rewrite_row_column_slice():
   }
   result = kthutils.forms.rewrite_row(row, rewriter)
   assert result == "Range: col1\tcol2"
+
+def test_rewrite_row_constant_value():
+  # A value substitution ignores the row entirely.
+  row = ["data", "Godkänt"]
+  rewriter = {
+    "format_string": "Passed: {passed}",
+    "substitutions": {
+      "passed": {"value": "Nej"}
+    }
+  }
+  result = kthutils.forms.rewrite_row(row, rewriter)
+  assert result == "Passed: Nej"
 @
 
 The lab week rewriter (\cref{LabWeekRewriter}) relies on a stepped slice and a
@@ -1605,10 +1677,22 @@ character.
 \item[regex] The regex to apply to the data in the column.
 This should have the form [[s/regex/replacement/]].
 The [[/]] can be any character, same as in [[sed]].
+\item[value] A fixed string to use as the substitution, ignoring the row.
+Format strings are shared between forms (\cref{LabWeekRewriter}), and a form
+may lack a column that the format string still refers to.
+Rather than dropping the field from the shared format string, or faking it
+with a regex that rewrites anything to a constant, the rewriter for that
+form supplies the value directly.
 \end{description}
+A [[value]] substitution needs neither column nor regex, so we handle it
+first and skip the rest of the loop body.
 <<create [[substitutions]] from [[row]] and [[rewriter]]>>=
 substitutions = {}
 for variable, var_subst in rewriter["substitutions"].items():
+  if "value" in var_subst:
+    substitutions[variable] = var_subst["value"]
+    continue
+
   if "column" in var_subst:
     <<make [[data]] contain the columns from [[row]]>>
   else: