Fine-tune the Community reports

cloudexit/engine.py

@@ -415,15 +415,19 @@ def perform_risk_assessment(exit_strategy: int, report_path: str) -> dict[str, A
         return {"success": False, "logs": str(e)}
 
 
-def _load_tfstate_scope(raw_data_path: str) -> dict[str, Any] | None:
-    # A missing or unreadable manifest must not fail report generation.
+def _load_tfstate_scope(
+    raw_data_path: str,
+) -> tuple[dict[str, Any] | None, dict[str, Any] | None]:
+    # A missing or unreadable manifest must not fail report generation. Older
+    # manifests have no coverage.
     manifest_path = os.path.join(raw_data_path, "tfstate_manifest.json")
     try:
         with open(manifest_path, "r", encoding="utf-8") as manifest_file:
-            return json.load(manifest_file).get("scope")
+            manifest = json.load(manifest_file)
+        return manifest.get("scope"), manifest.get("coverage")
     except (OSError, json.JSONDecodeError) as e:
         logger.warning("Could not read the tfstate manifest for report scope: %s", e)
-        return None
+        return None, None
 
 
 # Stage 8
@@ -477,10 +481,10 @@ def generate_report(
         reports = {}
 
         # One context feeds all three reports, so they say the same things
-        tfstate_scope = (
+        tfstate_scope, tfstate_coverage = (
             _load_tfstate_scope(raw_data_path)
             if provider_details.get("tfstatePath")
-            else None
+            else (None, None)
         )
         provenance = collect_provenance(db_path, run_id=sync_id, upload=upload)
         context = build_context(
@@ -496,6 +500,7 @@ def generate_report(
             alternative_technology_organizations=alternative_technology_organizations,
             exit_strategy=exit_strategy,
             tfstate_scope=tfstate_scope,
+            tfstate_coverage=tfstate_coverage,
             provenance=provenance,
             report_path=report_path,
         )

cloudexit/platform/upload.py

@@ -250,7 +250,7 @@ def read_sync_files(
     return found, problems
 
 
-REPORT_KEYS = ("status", "reason", "sync_id", "assessment_id", "url")
+REPORT_KEYS = ("status", "reason", "sync_id", "assessment_id", "url", "spooled")
 
 
 def _record(result: UploadResult | None) -> dict[str, Any] | None:
@@ -273,9 +273,9 @@ def _record(result: UploadResult | None) -> dict[str, Any] | None:
     return {key: value for key, value in info.items() if value}
 
 
-def report_info(result: UploadResult | None) -> dict[str, str] | None:
-    # What the local report may show: the status and the link, nothing the
-    # platform computed.
+def report_info(result: UploadResult | None) -> dict[str, Any] | None:
+    # What the local report may show: the status, the link and whether a failed
+    # run was kept, nothing the platform computed.
     record = _record(result)
     if record is None:
         return None

cloudexit/report/context.py

@@ -38,8 +38,10 @@ class Provenance:
     uploaded: bool = False
     platform_url: str | None = None
     assessment_id: str | None = None
-    # Why a requested upload did not happen, when it did not
+    # A requested upload that failed, why, and whether the run was kept
+    upload_failed: bool = False
     upload_reason: str | None = None
+    upload_spooled: bool = False
 
     @property
     def run_short_id(self):
@@ -79,9 +81,11 @@ def collect_provenance(db_path, *, run_id=None, upload=None, collected_at=None):
         uploaded=upload.get("status") == "uploaded",
         platform_url=upload.get("url") if upload.get("status") == "uploaded" else None,
         assessment_id=upload.get("assessment_id"),
+        upload_failed=upload.get("status") == "failed",
         upload_reason=(
-            upload.get("reason") if upload.get("status") != "uploaded" else None
+            upload.get("reason") if upload.get("status") == "failed" else None
         ),
+        upload_spooled=bool(upload.get("spooled")) and upload.get("status") == "failed",
     )
 
 
@@ -147,6 +151,14 @@ class ScopeFacts:
     resource_group: str = ""
     file_name: str = ""
     locations: tuple = ()
+    # State file instances left out: of other providers, of types not in the
+    # data catalogue
+    not_assessed_other: int = 0
+    not_assessed_unmapped: int = 0
+
+    @property
+    def not_assessed(self):
+        return self.not_assessed_other + self.not_assessed_unmapped
 
 
 @dataclass
@@ -316,7 +328,14 @@ def _alternative_entries(
     return entries
 
 
-def scope_facts(provider_details, cloud_service_provider, tfstate_scope):
+def _excluded(coverage, key):
+    value = (coverage or {}).get(key)
+    return value if isinstance(value, int) and value > 0 else 0
+
+
+def scope_facts(
+    provider_details, cloud_service_provider, tfstate_scope, tfstate_coverage=None
+):
     provider = PROVIDER_NAMES.get(cloud_service_provider, "–")
     if provider_details.get("tfstatePath"):
         scope = tfstate_scope or {
@@ -343,6 +362,12 @@ def scope_facts(provider_details, cloud_service_provider, tfstate_scope):
             tuple(rows),
             file_name=scope.get("file") or "",
             locations=locations,
+            not_assessed_other=_excluded(
+                tfstate_coverage, "instances_excluded_other_provider"
+            ),
+            not_assessed_unmapped=_excluded(
+                tfstate_coverage, "instances_excluded_unmapped"
+            ),
         )
     if cloud_service_provider == 1:
         rows = (
@@ -379,6 +404,7 @@ def build_context(
     alternative_technology_organizations: list[dict[str, Any]],
     exit_strategy: int,
     tfstate_scope: dict[str, Any] | None,
+    tfstate_coverage: dict[str, Any] | None = None,
     provenance: Provenance,
     report_path: str,
 ) -> ReportContext:
@@ -392,7 +418,7 @@ def build_context(
         provider=PROVIDER_NAMES.get(csp, "–"),
         exit_strategy=exit_strategy,
         strategy_code=STRATEGY_CODES.get(int(exit_strategy), ""),
-        scope=scope_facts(provider_details, csp, tfstate_scope),
+        scope=scope_facts(provider_details, csp, tfstate_scope, tfstate_coverage),
         provenance=provenance,
         resources=resources,
         costs=_cost_entries(cost_data, provenance.collected_at),

cloudexit/report/evidence.py

@@ -53,6 +53,22 @@ def build_document_table(
     return key_value_table(rows, context.appearance)
 
 
+def _not_assessed_text(scope, b):
+    parts = [
+        _counted(b, key, n)
+        for key, n in (
+            ("ev_not_assessed_other", scope.not_assessed_other),
+            ("ev_not_assessed_unmapped", scope.not_assessed_unmapped),
+        )
+        if n
+    ]
+    return ", ".join(parts)
+
+
+def _counted(b, key, n):
+    return b[f"{key}_one"] if n == 1 else b[f"{key}_many"].format(n=n)
+
+
 def scope_rows(context, b):
     # (label, text, monospaced), the same for the PDF and the HTML report
     scope = context.scope
@@ -91,6 +107,8 @@ def scope_rows(context, b):
         ):
             if key in values:
                 rows.append((b[label], values[key] or b["ev_not_determinable"], False))
+        if scope.not_assessed:
+            rows.append((b["ev_not_assessed"], _not_assessed_text(scope, b), False))
     source = b["ev_source_tfstate" if scope.kind == "tfstate" else "ev_source_live"]
     rows.append(
         (
@@ -117,6 +135,19 @@ def link_markup(url, color):
     return f'<a href={quoteattr(url)} color="{color}">{escape(url)}</a>'
 
 
+def platform_failure(provenance, b):
+    # What happened to an upload that failed, as it was true when the report was written
+    if provenance.upload_reason:
+        text = b["ev_platform_failed"].format(
+            reason=provenance.upload_reason.rstrip(". ")
+        )
+    else:
+        text = b["ev_platform_failed_no_reason"]
+    if provenance.upload_spooled:
+        text = f"{text} {b['ev_platform_kept']}"
+    return text
+
+
 def build_provenance_table(context, b):
     provenance = context.provenance
     if provenance.catalogue_version and provenance.catalogue_sha256:
@@ -137,7 +168,11 @@ def build_provenance_table(context, b):
         text = escape(b["ev_platform_uploaded"])
         if provenance.platform_url:
             text = f"{text}: {link_markup(provenance.platform_url, context.appearance.heading2_color)}"
+        elif provenance.assessment_id:
+            text = f"{text} · {escape(b['fa_uploaded_id'].format(assessment_id=provenance.assessment_id))}"
         rows.append((b["ev_platform_label"], Paragraph(text, KV_VALUE)))
+    elif provenance.upload_failed:
+        rows.append((b["ev_platform_label"], platform_failure(provenance, b)))
     else:
         rows.append((b["ev_platform_label"], b["ev_platform_none"]))
     return key_value_table(rows, context.appearance)

cloudexit/report/executive_summary.py

@@ -14,7 +14,6 @@
 from .sections import (
     alternative_technologies,
     costs_overview,
-    full_assessment,
     glossary,
     introduction,
     resource_inventory,
@@ -45,8 +44,6 @@ def build_content(context, styles, b):
     content.append(Spacer(1, 18))
     content += glossary(context, styles, b)
     content.append(chapter_break())
-    content += full_assessment(context, styles, b)
-    content.append(chapter_break())
     content += legal_notices(context, styles, b)
     return content
 

cloudexit/report/html_sections.py

@@ -1,6 +1,6 @@
 from markupsafe import Markup
 
-from .evidence import scope_rows
+from .evidence import platform_failure, scope_rows
 from .helpers import (
     format_change,
     initials,
@@ -11,12 +11,9 @@
     strategy_name,
 )
 from .html_egress import egress_page
-from .notices import text_parts
+from .notices import icon_notice, text_parts
 from .palette import SEVERITIES
 from .sections import (
-    AWS_ICONS_URL,
-    AWS_TRADEMARK_URL,
-    AZURE_ICONS_URL,
     PRODUCT_SITES,
     REPOSITORY_URL,
 )
@@ -406,6 +403,15 @@ def _inventory(context, b):
         ]
         if scope.kind == "tfstate":
             caption_parts.append(b["html_inv_not_managed"])
+    if scope.not_assessed:
+        key = (
+            "html_inv_not_assessed_one"
+            if scope.not_assessed == 1
+            else "html_inv_not_assessed_many"
+        )
+        caption_parts.append(
+            b[key].format(n=scope.not_assessed, about=b["html_nav_about"])
+        )
 
     if scope.kind == "azure":
         empty = b["html_inv_empty_resource_group"].format(
@@ -544,10 +550,8 @@ def _count_more(b, n):
 def _platform_cell(context, b):
     provenance = context.provenance
     if not provenance.uploaded:
-        if provenance.upload_reason:
-            return [
-                b["html_platform_not_uploaded"].format(reason=provenance.upload_reason)
-            ]
+        if provenance.upload_failed:
+            return [platform_failure(provenance, b)]
         return [b["ev_platform_none"]]
     parts = [b["ev_platform_uploaded"]]
     if provenance.platform_url:
@@ -667,20 +671,7 @@ def _about(context, egress, b):
 
 def _notices(context, egress, b):
     provenance = context.provenance
-    icons = None
-    if context.cloud_service_provider == 2:
-        icons = text_parts(
-            b["notices_icons_aws"],
-            {
-                "icons_url": (b["html_label_aws_icons"], AWS_ICONS_URL),
-                "guidelines_url": (b["html_label_aws_guidelines"], AWS_TRADEMARK_URL),
-            },
-        )
-    elif context.cloud_service_provider == 1:
-        icons = text_parts(
-            b["notices_icons_azure"],
-            {"icons_url": (b["html_label_azure_icons"], AZURE_ICONS_URL)},
-        )
+    icons = icon_notice(context.cloud_service_provider, b)
     if context.scope.kind == "tfstate":
         figures = b["html_figures_tfstate"]
     else:

cloudexit/report/notices.py

@@ -37,14 +37,22 @@ def text_parts(template, links, values=None):
     return parts
 
 
-def _icon_notice(provider, b):
+def icon_notice(provider, b):
+    # The names are the links: the HTML shows the name, the PDF the name and, in
+    # brackets, the address
     if provider == 2:
         return text_parts(
             b["notices_icons_aws"],
-            {"icons_url": AWS_ICONS_URL, "guidelines_url": AWS_TRADEMARK_URL},
+            {
+                "icons": (b["notices_label_aws_icons"], AWS_ICONS_URL),
+                "guidelines": (b["notices_label_aws_guidelines"], AWS_TRADEMARK_URL),
+            },
         )
     if provider == 1:
-        return text_parts(b["notices_icons_azure"], {"icons_url": AZURE_ICONS_URL})
+        return text_parts(
+            b["notices_icons_azure"],
+            {"icons": (b["notices_label_azure_icons"], AZURE_ICONS_URL)},
+        )
     return None
 
 
@@ -60,7 +68,7 @@ def notices_html(provider, generated_at, b=BLOCKS, egress=False):
         *([[b["notices_egress"]]] if egress else []),
     ]
     trademarks = [[b["notices_trademark_own"]], [b["notices_trademark_others"]]]
-    icons = _icon_notice(provider, b)
+    icons = icon_notice(provider, b)
     if icons:
         trademarks.append(icons)
     return {
@@ -85,9 +93,20 @@ def notices_html(provider, generated_at, b=BLOCKS, egress=False):
     }
 
 
+def _link_markup(label, url, color):
+    # A link that is not its own address prints the address too, for print
+    if label == url:
+        return link_markup(url, color)
+    return f"{escape(label)} ({link_markup(url, color)})"
+
+
 def parts_markup(parts, color):
     return "".join(
-        link_markup(part[1], color) if isinstance(part, tuple) else escape(part)
+        (
+            _link_markup(part[0], part[1], color)
+            if isinstance(part, tuple)
+            else escape(part)
+        )
         for part in parts
     )