TheAnachronism/docspell
1
0
Fork 0
mirror of https://github.com/TheAnachronism/docspell.git synced 2025-06-21 18:08:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

Set item due date candidate

Browse Source 
After processing, set the due date of an item to the first candidate.
The earliest due date is considered best match.
This commit is contained in:
Eike Kettner
2020-03-20 22:38:03 +01:00
parent fd48dace9d
commit cba466ed47
4 changed files with 79 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files

71
modules/joex/src/main/scala/docspell/joex/process/EvalProposals.scala
View File

@ -1,5 +1,6 @@
package docspell.joex.process
import java.time.{LocalDate, Period}
import cats.implicits._
import cats.effect.Sync
import docspell.common._
@ -12,54 +13,74 @@ object EvalProposals {
def apply[F[_]: Sync](data: ItemData): Task[F, ProcessItemArgs, ItemData] =
Task { _ =>
val metas = data.metas.map(reorderCandidates)
data.copy(metas = metas).pure[F]
Timestamp
.current[F]
.map { now =>
val metas = data.metas.map(reorderCandidates(now.toUtcDate))
data.copy(metas = metas)
}
}
def reorderCandidates(rm: RAttachmentMeta): RAttachmentMeta = {
def reorderCandidates(now: LocalDate)(rm: RAttachmentMeta): RAttachmentMeta = {
val list = rm.proposals.getTypes.toList
.map(mpt => rm.proposals.find(mpt) match {
case Some(mp) =>
val v = mp.values.sortBy(weight(rm, mp))
Some(mp.copy(values = v))
case None =>
None
})
.map(mpt =>
rm.proposals.find(mpt) match {
case Some(mp) =>
val v = mp.values.sortBy(weight(rm, mp, now))
Some(mp.copy(values = v))
case None =>
None
}
)
rm.copy(proposals = MetaProposalList(list.flatMap(identity)))
}
def weight(rm: RAttachmentMeta, mp: MetaProposal)(cand: MetaProposal.Candidate): Double = {
val textLen = rm.content.map(_.length).getOrElse(0)
val tagCount = cand.origin.size.toDouble
val pos = cand.origin.map(_.startPosition).min
val words = cand.origin.map(_.label.split(' ').length).max.toDouble
val nerFac = cand.origin.map(label => nerTagFactor(label.tag, mp.proposalType)).min
(1 / words) * (1 / tagCount) * positionWeight(pos, textLen) * nerFac
}
def weight(rm: RAttachmentMeta, mp: MetaProposal, ref: LocalDate)(
cand: MetaProposal.Candidate
): Double =
mp.proposalType match {
case MetaProposalType.DueDate =>
//for due dates, sort earliest on top
MetaProposal
.parseDate(cand)
.map { ld =>
val p = Period.between(ref, ld)
// conversion only for sorting
val d = p.getYears * 365 + p.getMonths * 31 + p.getDays
d.toDouble
}
.getOrElse(2000.0)
case _ =>
val textLen = rm.content.map(_.length).getOrElse(0)
val tagCount = cand.origin.size.toDouble
val pos = cand.origin.map(_.startPosition).min
val words = cand.origin.map(_.label.split(' ').length).max.toDouble
val nerFac =
cand.origin.map(label => nerTagFactor(label.tag, mp.proposalType)).min
(1 / words) * (1 / tagCount) * positionWeight(pos, textLen) * nerFac
}
def positionWeight(pos: Int, total: Int): Double = {
def positionWeight(pos: Int, total: Int): Double =
if (total <= 0) 1
else {
val p = math.abs(pos.toDouble / total.toDouble)
if (p < 0.7) p / 2
else p
}
}
def nerTagFactor(tag: NerTag, mt: MetaProposalType): Double =
tag match {
case NerTag.Date => 1.0
case NerTag.Email => 0.5
case NerTag.Date => 1.0
case NerTag.Email => 0.5
case NerTag.Location => 1.0
case NerTag.Misc => 1.0
case NerTag.Misc => 1.0
case NerTag.Organization =>
if (mt == MetaProposalType.CorrOrg) 0.8
else 1.0
case NerTag.Person =>
if (mt == MetaProposalType.CorrPerson ||
mt == MetaProposalType.ConcPerson) 0.8
mt == MetaProposalType.ConcPerson) 0.8
else 1.0
case NerTag.Website => 0.5
}

28
modules/joex/src/main/scala/docspell/joex/process/LinkProposal.scala
View File

@ -50,20 +50,38 @@ object LinkProposal {
mpt match {
case MetaProposalType.CorrOrg =>
ctx.logger.debug(s"Updating item organization with: ${value.id}") *>
ctx.store.transact(RItem.updateCorrOrg(itemId, ctx.args.meta.collective, Some(value)))
ctx.store.transact(
RItem.updateCorrOrg(itemId, ctx.args.meta.collective, Some(value))
)
case MetaProposalType.ConcPerson =>
ctx.logger.debug(s"Updating item concerning person with: $value") *>
ctx.store.transact(RItem.updateConcPerson(itemId, ctx.args.meta.collective, Some(value)))
ctx.store.transact(
RItem.updateConcPerson(itemId, ctx.args.meta.collective, Some(value))
)
case MetaProposalType.CorrPerson =>
ctx.logger.debug(s"Updating item correspondent person with: $value") *>
ctx.store.transact(RItem.updateCorrPerson(itemId, ctx.args.meta.collective, Some(value)))
ctx.store.transact(
RItem.updateCorrPerson(itemId, ctx.args.meta.collective, Some(value))
)
case MetaProposalType.ConcEquip =>
ctx.logger.debug(s"Updating item concerning equipment with: $value") *>
ctx.store.transact(RItem.updateConcEquip(itemId, ctx.args.meta.collective, Some(value)))
ctx.store.transact(
RItem.updateConcEquip(itemId, ctx.args.meta.collective, Some(value))
)
case MetaProposalType.DocDate =>
ctx.logger.debug(s"Not linking document date suggestion ${value.id}").map(_ => 0)
case MetaProposalType.DueDate =>
ctx.logger.debug(s"Not linking document date suggestion ${value.id}").map(_ => 0)
MetaProposal.parseDate(value) match {
case Some(ld) =>
val ts = Timestamp.from(ld.atStartOfDay(Timestamp.UTC))
ctx.logger.debug(s"Updating item due-date suggestion ${value.id}") *>
ctx.store.transact(
RItem.updateDueDate(itemId, ctx.args.meta.collective, Some(ts))
)
case None =>
ctx.logger.info(s"Cannot read value '${value.id}' into a date.") *>
0.pure[F]
}
}
sealed trait Result {