ElixirForge/bom-local-service
1
0
Fork 0
mirror of https://github.com/alexhopeoconnor/bom-local-service.git synced 2026-10-03 21:21:59 +10:00
Code Releases 7 Activity

Release v0.0.6: resilient BOM metadata parsing.

Browse Source 
Read C07_WeatherMetadata via innerText, accept observation prefixes (a minute,
N minutes, an hour, N hours) and forecast N hours ago. Version 0.0.6.
This commit is contained in:
alex
2026-05-07 05:49:51 +10:00
parent 28e8cb4fd1
commit 8eecce00a6
2 changed files with 28 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files
BomLocalService.csproj
+1 -1
View File
@@ -4,7 +4,7 @@
<TargetFramework>net9.0</TargetFramework> <TargetFramework>net9.0</TargetFramework>
<Nullable>enable</Nullable> <Nullable>enable</Nullable>
<ImplicitUsings>enable</ImplicitUsings> <ImplicitUsings>enable</ImplicitUsings>
<Version>0.0.5</Version> <Version>0.0.6</Version>
</PropertyGroup> </PropertyGroup>
<ItemGroup> <ItemGroup>
Services/TimeParsingService.cs
+27 -19
View File
@@ -41,26 +41,30 @@ public class TimeParsingService : ITimeParsingService
{ {
try try
{ {
// Look for the weather metadata section using the correct selector // BOM layouts: legacy map used inner divs; location/spatial pages often use C07_WeatherMetadata with
// HTML structure: <section data-testid="weatherMetadata"> with aria-label="Last updated" // a single section and visible text in the section (no div split). Prefer normalized innerText of the
var metadataSection = page.Locator("section[data-testid='weatherMetadata'], section[aria-label='Last updated']").First; // section so both shapes parse the same.
var metadataSection = page.Locator(
// Extract text from each div separately to preserve structure "section[data-testid='weatherMetadata'], " +
// Structure: <div>Observations: X minutes ago, 9:40 pm AEST</div><div>at Gympie...</div><div>Forecast: ...</div> "section[aria-label='Last updated'], " +
"section[data-component='C07_WeatherMetadata']").First;
var lastUpdatedText = await page.EvaluateAsync<string>(@"() => { var lastUpdatedText = await page.EvaluateAsync<string>(@"() => {
const section = document.querySelector('section[data-testid=""weatherMetadata""]') || const section =
document.querySelector('section[aria-label=""Last updated""]'); document.querySelector('section[data-testid=""weatherMetadata""]') ||
document.querySelector('section[aria-label=""Last updated""]') ||
document.querySelector('section[data-component=""C07_WeatherMetadata""]');
if (!section) return null; if (!section) return null;
const raw = section.innerText || section.textContent || '';
// Get all divs and join with newlines to preserve structure return raw.replace(/\s+/g, ' ').trim();
const divs = section.querySelectorAll('div');
return Array.from(divs).map(div => div.textContent.trim()).filter(text => text).join(' ');
}"); }");
if (string.IsNullOrEmpty(lastUpdatedText)) if (string.IsNullOrEmpty(lastUpdatedText))
{ {
// Fallback: get all text content var fallback = await metadataSection.InnerTextAsync();
lastUpdatedText = await metadataSection.TextContentAsync(); lastUpdatedText = string.IsNullOrEmpty(fallback)
? null
: Regex.Replace(fallback, @"\s+", " ").Trim();
} }
return ParseLastUpdatedText(lastUpdatedText ?? string.Empty); return ParseLastUpdatedText(lastUpdatedText ?? string.Empty);
@@ -82,8 +86,10 @@ public class TimeParsingService : ITimeParsingService
_logger.LogInformation("Parsing last updated text: {Text}", text); _logger.LogInformation("Parsing last updated text: {Text}", text);
// Parse observation time from time string (e.g., "9:40 pm AEST") // Parse observation time from time string (e.g., "9:40 pm AEST")
// Ignore "minutes ago" - client will calculate this dynamically from UTC timestamp // Ignore relative prefix — BOM uses "a minute ago", "6 minutes ago", "an hour ago", "2 hours ago", etc.
var observationMatch = Regex.Match(text, @"Observations:\s*(?:\d+\s*minutes?\s*ago)?[,\s]+([\d:]+(?:\s*[ap]m)?)\s+([A-Z]{3,4})(?:at|\s+at|,|$)", RegexOptions.IgnoreCase); var observationMatch = Regex.Match(text,
@"Observations:\s*(?:a\s+minute\s+ago|\d+\s*minutes?\s+ago|an\s+hour\s+ago|\d+\s*hours?\s+ago)?[,\s]+([\d:]+(?:\s*[ap]m)?)\s+([A-Z]{3,4})(?:at|\s+at|,|$)",
RegexOptions.IgnoreCase);
if (observationMatch.Success && observationMatch.Groups.Count >= 3 && observationMatch.Groups[1].Success) if (observationMatch.Success && observationMatch.Groups.Count >= 3 && observationMatch.Groups[1].Success)
{ {
var timeStr = observationMatch.Groups[1].Value.Trim(); var timeStr = observationMatch.Groups[1].Value.Trim();
@@ -114,8 +120,10 @@ public class TimeParsingService : ITimeParsingService
} }
// Parse forecast time from time string (e.g., "7:50 pm AEST") // Parse forecast time from time string (e.g., "7:50 pm AEST")
// Ignore "minutes ago" - client will calculate this dynamically from UTC timestamp // Ignore relative prefix (minutes/hours ago) — client derives "ago" from the UTC timestamp
var forecastMatch = Regex.Match(text, @"Forecast:\s*(?:an\s+hour\s+ago|\d+\s*minutes?\s*ago)?[,\s]*([\d:]+(?:\s*[ap]m)?)\s+([A-Z]+)(?:\s+at|$)", RegexOptions.IgnoreCase); var forecastMatch = Regex.Match(text,
@"Forecast:\s*(?:\d+\s*hours?\s+ago|an\s+hour\s+ago|\d+\s*minutes?\s*ago)?[,\s]*([\d:]+(?:\s*[ap]m)?)\s+([A-Z]+)(?:\s+at|$)",
RegexOptions.IgnoreCase);
if (forecastMatch.Success && forecastMatch.Groups[1].Success) if (forecastMatch.Success && forecastMatch.Groups[1].Success)
{ {
var timeStr = forecastMatch.Groups[1].Value.Trim(); var timeStr = forecastMatch.Groups[1].Value.Trim();