Un aperçu des métadonnées de la presse numérisées

db = DuckDBClient.of({
fascicules: FileAttachment("./fascicules.parquet"),
periodiques: FileAttachment("./periodiques.parquet"),
})
theme = ({
  style: {fontSize: "16px"},
  locale:"fr",
  width: 800,
})
fr = d3.formatLocale({decimal: ',', thousands: ' ', grouping: [3], currency: ['€', '']})

Informations du catalogue général

viewof ark = Inputs.text({label: "Ark catalogue", placeholder: "Ark catalogue", value:"cb34355551z", submit:true})

{
  const data = await db.query(`
  SELECT 
    titre, ark_catalogue, type_date, periodicite,
    first_date, last_date,
    lieu_edition, classement
  FROM periodiques WHERE ark_catalogue = ?
  `, [ark])
//| label: tbl-catalogue
//| fig-cap: "Extrait des métadonnées issues du catalogue"
return Inputs.table(data)
}

Volumétrie des fascicules sur Gallica

{
const data = await db.query(`
SELECT 
  COUNT(*) AS Fascicules,
  COUNT(*) FILTER (WHERE meta.ocr_mean) AS 'Fascicules OCRisés',
  COUNT(*) FILTER (WHERE meta.intramuros) AS Intramuros,
  COUNT(*) FILTER (WHERE meta.retronews) AS Retronews,
  MIN(date.year) AS 'Date début',
  MAX(date.year) AS 'Date de fin',
FROM 
  fascicules 
WHERE
  ark_catalogue= ?
`, [ark])
return Inputs.table(data, {locale:"fr"})

}

{

const data = await db.query(`
SELECT 
  date.year AS Année,
  COUNT(*) AS Fascicules,
  COUNT(*) FILTER (WHERE meta.ocr_mean) AS 'Fascicules OCRisés',
  COUNT(*) FILTER (WHERE meta.intramuros) AS Intramuros,
  COUNT(*) FILTER (WHERE meta.retronews) AS Retronews,
FROM 
  fascicules 
WHERE
  ark_catalogue= ?
GROUP BY
  date.year
ORDER BY date.year
`, [ark])
//| label: fig-lacroix-ocr
//| fig-cap: "Répartition du taux d'OCR moyen pour le journal La Croix"
return Plot.plot({
  ...theme,
  grid: true,
  color: {
    legend: true,
    domain: ["#dbc257", "#db5f57"],
    range: ["#dbc257", "#db5f57"],
    tickFormat: (d) =>
      d === "#dbc257"
        ? "Fascicules OCRisés"
        : d === "#db5f57"
        ? "Fascicules"
        : "you shouldn't be here"
  },
  marks: [
  Plot.lineY(data,
    {
      x:"Année",
      y:"Fascicules OCRisés",
      curve: "catmull-rom",
      marker:"circle-stroke",
      tip:"x",
      stroke:"#dbc257"
    }
  ),
  Plot.lineY(data,
    {
      x:"Année",
      y:"Fascicules",
      curve: "catmull-rom",
      marker:"circle-stroke",
      tip:"x",
      stroke:"#db5f57",
      label: "Nombre de fascicules total"
    }),
  Plot.axisX(
    {
      anchor: "bottom",
      marginBottom:60,
      tickSpacing: 80,
      tickFormat: ".2d"
    }),
  ],
})
}

Qualité de l’OCR

{
const data = await db.query(`
SELECT 
  date.year,
  meta.ocr_mean
FROM 
  fascicules 
WHERE
  ark_catalogue= ?
  AND meta.ocr_mean
`, [ark])
return Plot.plot({
  ...theme,
  grid: true,
  x:{label:"OCR Moyen (%)"},
  y:{label:"Nombre de fascicules"},
  marks: [
    Plot.rectY(
      data,
      Plot.binX({y:"count"},{x: "ocr_mean",tip:"x"})),
      Plot.ruleY([0]),
  Plot.axisX(
    {
      anchor: "bottom",
      marginBottom:40,
      tickSpacing: 80,
    }),
  Plot.axisY(
    {
      marginLeft:50,
      tickSpacing: 80,
      tickFormat: ".2d",
    }),
  ]
})
}

{
const data = await db.query(`
SELECT 
  date.year,
  meta.ocr_mean
FROM 
  fascicules 
WHERE
  ark_catalogue= ?
  AND meta.ocr_mean
`, [ark])
return Plot.plot({
  ...theme,
  grid: true,
  color: { zero: true, scheme: "Cividis",legend:true},
  width:800,
  height:1500,
  facet:{
    data: data,
    marginLeft: 80,
    y: "year",
  },
  x:{label:"OCR Moyen (%)", axis:"both"},
  marks:[
  Plot.barX(
    data,
    Plot.binX({fill:"proportion-facet"},
      {
        x:"ocr_mean",
        tip:"x",
        sort:{x:"fill"}
      }
    )
  ),
  Plot.axisX(
    {
      anchor: "bottom",
      marginBottom:40,
      tickSpacing: 80,
    }),
  ]
})
}