freeharness 0.4.1

package/canonical/skills/report-formatting/scripts/pptx/pptx_office.py

@@ -0,0 +1,144 @@
+#!/usr/bin/env python3
+"""Self-contained pptx office ops for report-formatting (no Anthropic container
+dependency). A .pptx is an OOXML zip; this provides unpack/pack/clean/render.
+
+Usage:
+  python pptx_office.py unpack <in.pptx> <dir>
+  python pptx_office.py pack   <dir> <out.pptx>
+  python pptx_office.py clean  <dir>
+  python pptx_office.py render <in.pptx> [outdir]    # needs soffice + pdftoppm
+"""
+import os
+import re
+import sys
+import shutil
+import zipfile
+import subprocess
+
+
+def unpack(pptx, out_dir):
+    """Extract a .pptx (OOXML zip) into out_dir, replacing it if it exists.
+
+    Args:
+        pptx: Path to the source .pptx file.
+        out_dir: Directory to extract into; removed first if present.
+    """
+    if os.path.isdir(out_dir):
+        shutil.rmtree(out_dir)
+    os.makedirs(out_dir, exist_ok=True)
+    with zipfile.ZipFile(pptx) as z:
+        z.extractall(out_dir)
+    print(f"unpacked {pptx} -> {out_dir}")
+
+
+def pack(in_dir, out_pptx):
+    """Zip the OOXML tree in in_dir into out_pptx ([Content_Types].xml first).
+
+    Overwrites out_pptx if it exists.
+
+    Args:
+        in_dir: Directory holding the unpacked OOXML parts.
+        out_pptx: Destination .pptx path.
+    """
+    files = []
+    for root, _dirs, names in os.walk(in_dir):
+        for n in names:
+            full = os.path.join(root, n)
+            arc = os.path.relpath(full, in_dir).replace(os.sep, "/")
+            files.append((full, arc))
+    # [Content_Types].xml conventionally first.
+    files.sort(key=lambda fa: (fa[1] != "[Content_Types].xml", fa[1]))
+    if os.path.exists(out_pptx):
+        os.remove(out_pptx)
+    with zipfile.ZipFile(out_pptx, "w", zipfile.ZIP_DEFLATED) as z:
+        for full, arc in files:
+            z.write(full, arc)
+    print(f"packed {in_dir} -> {out_pptx} ({len(files)} parts)")
+
+
+def clean(in_dir):
+    """Remove slide parts not referenced by presentation.xml.rels.
+
+    Also removes their rels and matching notesSlides. Minimal orphan sweep;
+    deletes files from in_dir in place.
+
+    Args:
+        in_dir: Directory holding the unpacked OOXML parts.
+    """
+    rels = os.path.join(in_dir, "ppt", "_rels", "presentation.xml.rels")
+    if not os.path.isfile(rels):
+        print("no presentation.xml.rels; nothing to clean")
+        return
+    txt = open(rels, encoding="utf-8").read()
+    referenced = set(re.findall(r'Target="(slides/slide\d+\.xml)"', txt))
+    slides_dir = os.path.join(in_dir, "ppt", "slides")
+    removed = 0
+    if os.path.isdir(slides_dir):
+        for n in os.listdir(slides_dir):
+            m = re.match(r"slide\d+\.xml$", n)
+            if not m:
+                continue
+            if f"slides/{n}" not in referenced:
+                os.remove(os.path.join(slides_dir, n))
+                rel = os.path.join(slides_dir, "_rels", n + ".rels")
+                if os.path.isfile(rel):
+                    os.remove(rel)
+                removed += 1
+    print(f"clean: removed {removed} orphan slide part(s)")
+
+
+def render(pptx, outdir="."):
+    """Render a .pptx to per-page JPEGs (soffice -> pdf -> pdftoppm).
+
+    Best-effort QA; prints the generated JPEG paths.
+
+    Args:
+        pptx: Path to the source .pptx.
+        outdir: Output directory for the pdf and jpegs.
+
+    Raises:
+        subprocess.CalledProcessError: If soffice or pdftoppm fails.
+    """
+    os.makedirs(outdir, exist_ok=True)
+    subprocess.run(
+        ["soffice", "--headless", "--convert-to", "pdf", "--outdir", outdir, pptx],
+        check=True,
+    )
+    pdf = os.path.join(outdir, os.path.splitext(os.path.basename(pptx))[0] + ".pdf")
+    subprocess.run(
+        ["pdftoppm", "-jpeg", "-r", "130", pdf, os.path.join(outdir, "slide")],
+        check=True,
+    )
+    jpegs = sorted(f for f in os.listdir(outdir) if f.startswith("slide") and f.endswith(".jpg"))
+    print("\n".join(os.path.join(outdir, j) for j in jpegs))
+
+
+def main(argv):
+    """Dispatch the CLI subcommand and return a process exit code.
+
+    Args:
+        argv: Argument vector (argv[1] is the subcommand).
+
+    Returns:
+        0 on success, 1 on usage error.
+    """
+    if len(argv) < 2:
+        print(__doc__)
+        return 1
+    cmd = argv[1]
+    if cmd == "unpack" and len(argv) == 4:
+        unpack(argv[2], argv[3])
+    elif cmd == "pack" and len(argv) == 4:
+        pack(argv[2], argv[3])
+    elif cmd == "clean" and len(argv) == 3:
+        clean(argv[2])
+    elif cmd == "render" and len(argv) in (3, 4):
+        render(argv[2], argv[3] if len(argv) == 4 else ".")
+    else:
+        print(__doc__)
+        return 1
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main(sys.argv))

package/canonical/skills/report-formatting/scripts/pptx/requirements.txt

@@ -0,0 +1,7 @@
+# Python deps for the .pptx track of report-formatting.
+# Install: pip install -r requirements.txt --break-system-packages
+# System tools (not pip): soffice (LibreOffice headless) + pdftoppm (poppler) for QA render.
+python-pptx>=1.0.0
+lxml>=5.0.0
+Pillow>=10.0.0
+defusedxml>=0.7.1

package/canonical/skills/report-formatting/scripts/pptx/status_report_blocks.py

@@ -0,0 +1,193 @@
+#!/usr/bin/env python3
+"""Reusable status-report slide builders (Claro theme), for the report-formatter
+agent. Generic — no project-specific data. Import alongside theme_claro_pptx:
+
+    import theme_claro_pptx as T
+    import status_report_blocks as B
+    from pptx import Presentation
+
+    prs = Presentation("deck.pptx")           # or Presentation() for a new deck
+    B.build_heatmap(prs.slides[i], title=..., rows=...)   # clears + rebuilds slide i
+    B.build_gantt(prs.slides[j], title=..., activities=...)
+
+Both functions CLEAR the target slide and rebuild it, so they work for new blank
+slides and for in-place rebuilds. Tested against the STT status report (Jun/2026).
+
+Two blocks:
+- build_heatmap: período × estágio status grid (Concluído/Em Execução/Pendente/
+  Parcial/Backlog), optional trailing data column (e.g. "Backlog (TIDs)").
+- build_gantt: Claro "Cronograma / Próximos Passos" — red header bar, dark week
+  columns, per-row accent + category chip + priority bar, category legend.
+"""
+import os
+import sys
+
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+import theme_claro_pptx as T  # noqa: E402
+from pptx.util import Inches  # noqa: E402
+from pptx.dml.color import RGBColor  # noqa: E402
+from pptx.enum.text import PP_ALIGN, MSO_ANCHOR  # noqa: E402
+
+# Status palette for the heatmap (key -> fill, icon, label).
+PARCIAL = RGBColor(0xD4, 0xA0, 0x17)  # gold — partial / incomplete data
+STATUS = {
+    "done": (T.STATUS_DONE, "✓", "Concluído"),
+    "inprogress": (T.STATUS_INPROGRESS, "▶", "Em Execução"),
+    "pending": (T.STATUS_PENDING, "○", "Pendente"),
+    "parcial": (PARCIAL, "◑", "Parcial"),
+    "backlog": (T.STATUS_BACKLOG, "–", "Backlog"),
+}
+FULL_LEGEND = ["done", "inprogress", "pending", "parcial", "backlog"]
+DEFAULT_STAGES = ["Download Áudio", "Transcrição", "Carga LakeHouse", "Validação"]
+PRIO_DISPLAY = {"Alta": "Prioridade Alta", "Média": "Prioridade Média", "Baixa": "Prioridade Baixa"}
+
+
+def build_heatmap(slide, *, title, subtitle, banner, rows, stages=None,
+                  backlog_col=True, data_header="Backlog (TIDs)",
+                  legend_keys=None, page=None, total=None):
+    """Build a período × estágio status heatmap; clears and rebuilds `slide`.
+
+    Args:
+        slide: Target slide; cleared and rebuilt in place.
+        title: Header title.
+        subtitle: Header subtitle.
+        banner: Banner strip text below the header.
+        rows: List of (label, [status_key per stage], data_str|None); data_str
+            fills the optional trailing column when backlog_col is True.
+        stages: Column captions (default: the four STT stages).
+        backlog_col: If True, add a trailing data column.
+        data_header: Caption for the trailing data column.
+        legend_keys: Subset/order of STATUS keys to show (default: FULL_LEGEND).
+        page: Footer page number, or None to omit the footer.
+        total: Footer total page count (defaults to `page`).
+    """
+    stages = stages or DEFAULT_STAGES
+    legend_keys = legend_keys or FULL_LEGEND
+    T.clear_slide_shapes(slide)
+    T.add_page_header(slide, title, subtitle)
+    T.add_rect(slide, Inches(0.45), Inches(1.32), Inches(12.4), Inches(0.4), T.VERY_LIGHT_GRAY)
+    T.add_text(slide, Inches(0.6), Inches(1.36), Inches(12.1), Inches(0.32), banner,
+               size=11, bold=True, color=T.CLARO_RED_DARK)
+    n = len(rows)
+    if backlog_col:
+        x_lab, w_lab = Inches(0.45), Inches(1.55)
+        x0, cw = Inches(2.05), Inches(2.0)
+        x_bk, w_bk = x0 + cw * len(stages) + Inches(0.05), Inches(2.45)
+    else:
+        x_lab, w_lab = Inches(0.45), Inches(1.9)
+        x0 = Inches(2.45)
+        cw = (Inches(12.85) - x0) / len(stages)
+        x_bk = None
+    y_hdr = Inches(1.95)
+    ch = Inches(0.45) if n <= 8 else Inches(0.33)
+    gap = Inches(0.08) if n <= 8 else Inches(0.035)
+    T.add_text(slide, x_lab, y_hdr, w_lab, Inches(0.3), "Período", size=10, bold=True, color=T.MID_GRAY)
+    for j, st in enumerate(stages):
+        T.add_text(slide, x0 + cw * j, y_hdr, cw, Inches(0.3), st, size=10, bold=True,
+                   color=T.MID_GRAY, align=PP_ALIGN.CENTER)
+    if backlog_col:
+        T.add_text(slide, x_bk, y_hdr, w_bk, Inches(0.3), data_header, size=10,
+                   bold=True, color=T.MID_GRAY, align=PP_ALIGN.CENTER)
+    y0 = y_hdr + Inches(0.35)
+    for i, (label, statuses, extra) in enumerate(rows):
+        y = y0 + i * (ch + gap)
+        T.add_rect(slide, x_lab, y, w_lab, ch, T.VERY_LIGHT_GRAY)
+        T.add_text(slide, x_lab + Inches(0.1), y, w_lab - Inches(0.1), ch, label,
+                   size=10, bold=True, color=T.DARK_GRAY, anchor=MSO_ANCHOR.MIDDLE)
+        for j, key in enumerate(statuses):
+            fill, icon, _ = STATUS[key]
+            T.add_filled_text(slide, x0 + cw * j, y, cw - Inches(0.1), ch, icon, fill=fill, size=13)
+        if backlog_col:
+            T.add_rect(slide, x_bk, y, w_bk, ch, T.LIGHT_GRAY)
+            T.add_text(slide, x_bk, y, w_bk, ch, extra or "—", size=11, bold=True,
+                       color=T.DARK_GRAY, align=PP_ALIGN.CENTER, anchor=MSO_ANCHOR.MIDDLE)
+    ly = y0 + n * (ch + gap) + Inches(0.12)
+    T.add_text(slide, Inches(0.45), ly, Inches(1.0), Inches(0.3), "Legenda:", size=9,
+               bold=True, color=T.MID_GRAY)
+    lx = Inches(0.45) + Inches(1.0)
+    for key in legend_keys:
+        fill, icon, lab = STATUS[key]
+        T.add_filled_text(slide, lx, ly, Inches(0.32), Inches(0.28), icon, fill=fill, size=10)
+        T.add_text(slide, lx + Inches(0.36), ly, Inches(2.0), Inches(0.3), lab, size=9,
+                   color=T.DARK_GRAY, anchor=MSO_ANCHOR.MIDDLE)
+        lx = lx + Inches(0.36) + Inches(min(len(lab) * 0.09 + 0.3, 2.0))
+    if page is not None:
+        T.add_page_footer(slide, page, total or page)
+
+
+def build_gantt(slide, *, title, subtitle, activities, week_labels,
+                page=None, total=None, header_left="Atividade / Frente"):
+    """Build the Claro 'Cronograma / Próximos Passos' gantt on `slide`.
+
+    Clears and rebuilds the slide in place.
+
+    Args:
+        slide: Target slide; cleared and rebuilt in place.
+        title: Header title.
+        subtitle: Header subtitle.
+        activities: List of {name, cat, start, end, prio}; `cat` is a key in
+            T.CAT_COLOR, start/end are 1-based week indices, prio in {Alta,
+            Média, Baixa}.
+        week_labels: One label per week column (use "Mes\\nSn").
+        page: Footer page number, or None to omit the footer.
+        total: Footer total page count (defaults to `page`).
+        header_left: Caption for the top-left header cell.
+    """
+    n_weeks = len(week_labels)
+    HDR_X, HDR_Y, HDR_W, HDR_H = Inches(0.45), Inches(1.55), Inches(4.2), Inches(0.48)
+    ACT_X, ACT_W = Inches(0.6), Inches(2.8)
+    CAT_X, CAT_W, CAT_H = Inches(3.4), Inches(1.15), Inches(0.24)
+    ACC_X, ACC_W = Inches(0.45), Inches(0.08)
+    GRID_X = Inches(4.65)
+    GRID_W = Inches(13.333) - GRID_X - Inches(0.45)
+    COL_W = GRID_W / n_weeks
+    ROW_Y0 = HDR_Y + HDR_H + Inches(0.04)
+    ROW_H, ROW_GAP = Inches(0.38), Inches(0.07)
+    BAR_H = Inches(0.22)
+    BAR_PADY = (ROW_H - BAR_H) / 2
+
+    T.clear_slide_shapes(slide)
+    T.add_page_header(slide, title, subtitle)
+    T.add_filled_text(slide, HDR_X, HDR_Y, HDR_W, HDR_H, header_left,
+                      fill=T.CLARO_RED, tcolor=T.WHITE, size=10, bold=True,
+                      align=PP_ALIGN.LEFT, anchor=MSO_ANCHOR.MIDDLE)
+    for w in range(n_weeks):
+        cx = GRID_X + COL_W * w
+        T.add_filled_text(slide, cx, HDR_Y, COL_W - Inches(0.01), HDR_H, week_labels[w],
+                          fill=T.DARK_GRAY, tcolor=T.WHITE, size=9, bold=True,
+                          align=PP_ALIGN.CENTER, anchor=MSO_ANCHOR.MIDDLE)
+    for i, a in enumerate(activities):
+        y = ROW_Y0 + i * (ROW_H + ROW_GAP)
+        color = T.CAT_COLOR[a["cat"]]
+        for w in range(n_weeks):
+            cx = GRID_X + COL_W * w
+            T.add_rect(slide, cx, y, COL_W - Inches(0.01), ROW_H, T.VERY_LIGHT_GRAY)
+        T.add_rect(slide, HDR_X, y, HDR_W, ROW_H, T.WHITE)
+        T.add_rect(slide, ACC_X, y, ACC_W, ROW_H, color)
+        T.add_text(slide, ACT_X, y, ACT_W, ROW_H, a["name"], size=7.5,
+                   color=T.DARK_GRAY, anchor=MSO_ANCHOR.MIDDLE)
+        chip_y = y + (ROW_H - CAT_H) / 2
+        T.add_filled_text(slide, CAT_X, chip_y, CAT_W, CAT_H, a["cat"], fill=color,
+                          tcolor=T.WHITE, size=7.5, bold=True, align=PP_ALIGN.CENTER,
+                          anchor=MSO_ANCHOR.MIDDLE, corner=True)
+        bx = GRID_X + COL_W * (a["start"] - 1) + Inches(0.03)
+        bw = COL_W * (a["end"] - a["start"] + 1) - Inches(0.06)
+        T.add_filled_text(slide, bx, y + BAR_PADY, bw, BAR_H,
+                          PRIO_DISPLAY.get(a["prio"], a["prio"]), fill=color,
+                          tcolor=T.WHITE, size=7.5, bold=True, align=PP_ALIGN.CENTER,
+                          anchor=MSO_ANCHOR.MIDDLE, corner=True)
+    ly = ROW_Y0 + len(activities) * (ROW_H + ROW_GAP) + Inches(0.14)
+    T.add_text(slide, Inches(0.45), ly, Inches(1.4), Inches(0.3), "Categorias:",
+               size=9, bold=True, color=T.MID_GRAY)
+    lx = Inches(1.5)
+    seen = []
+    for a in activities:
+        if a["cat"] not in seen:
+            seen.append(a["cat"])
+    for cat in seen:
+        T.add_rect(slide, lx, ly + Inches(0.05), Inches(0.2), Inches(0.2), T.CAT_COLOR[cat])
+        T.add_text(slide, lx + Inches(0.25), ly, Inches(1.6), Inches(0.3), cat, size=9,
+                   color=T.DARK_GRAY, anchor=MSO_ANCHOR.MIDDLE)
+        lx = lx + Inches(0.25 + min(len(cat) * 0.085 + 0.22, 1.6))
+    if page is not None:
+        T.add_page_footer(slide, page, total or page)